{ "cells": [ { "cell_type": "markdown", "id": "b96e984a-a9bc-4d29-8ae4-250cb7a7f65f", "metadata": {}, "source": [ "## Tutorial A4: Seqlets\n", "\n", "> deep_lift_shap/saturation mutagenesis\n", "\n", "Ultimately, our goal is to better understand the genomic sequences driving biological activity. To do this, we train machine learning models to make accurate predictions and then we use attribution methods (DeepLIFT/SHAP, in silico saturation mutagenesis, whatever) to highlight the individual nucleotides driving those model predictions. However, simply assigning some weight to each nucleotide is not particularly informative, nor is it a scalable way to understand what a model has learned.\n", "\n", "Accordingly, after calculating attribution scores, a common task is to identify the contiguous *spans* of high attribution characters that drive model predictions and, potentially, form protein binding sites. These spans are called \"seqlets\" and form the basis of subsequent algorithms that aim to distill what these machine learning methods have learned. For example, seqlets can be clustered into recurring patterns that drive model predictions (e.g., as TF-MoDISco does), but seqlets can also be used as an automated way of annotating sequences and extracting spatial relationships between motifs. At a high level, one can think of a seqlet as a base cis-regulatory unit identified by the model.\n", "\n", "Like with our other tutorials, let's begin by loading the Beluga model." ] }, { "cell_type": "code", "execution_count": 1, "id": "946c4c51-acd4-4d71-8ab2-12d8170c35a1", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 1, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import torch\n", "from model import Beluga\n", "\n", "model = Beluga()\n", "model.load_state_dict(torch.load(\"deepsea.beluga.torch\"))" ] }, { "cell_type": "markdown", "id": "deb0e15c-d936-4cf9-b0db-aae24f0c08e9", "metadata": {}, "source": [ "To demonstrate seqlet calling, we need clean attribution scores. In practice, one will usually run a seqlet caller on attributions from real genomic sequences. However, for demonstration purposes, we will randomly generate a sequence and insert three known motifs into it. Specifically, we will insert two identical AP-1 motifs and a YY-1 motif, at equal spacings from each other. Then, we will run DeepLIFT/SHAP using a DNase target." ] }, { "cell_type": "code", "execution_count": 2, "id": "3c17207c-6268-434c-b48e-7fd78611f49b", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "from matplotlib import pyplot as plt\n", "import seaborn; seaborn.set_style('whitegrid')\n", "\n", "from tangermeme.utils import random_one_hot\n", "from tangermeme.ersatz import multisubstitute\n", "from tangermeme.deep_lift_shap import deep_lift_shap\n", "from tangermeme.plot import plot_logo\n", "\n", "X = random_one_hot((1, 4, 2000), random_state=0).type(torch.float32) \n", "X = multisubstitute(X, [\"GTGACTCATC\", \"GTGACTCATC\", \"ACTTCCGCG\"], [25, 25])\n", "X_attr = deep_lift_shap(model, X, device='cpu', n_shuffles=20, target=57, random_state=0)\n", "\n", "plt.figure(figsize=(12, 2))\n", "ax = plt.subplot(111)\n", "plot_logo(X_attr[0, :, 925:1075], ax=ax)\n", "plt.ylim(-0.025, 0.25)\n", "\n", "plt.xlabel(\"Genomic Coordinate\")\n", "plt.ylabel(\"Attribution\")\n", "plt.title(\"DeepLIFT/SHAP of Marginalization\")\n", "plt.show()" ] }, { "cell_type": "markdown", "id": "88d2f6a6-09c0-4511-ba73-5c21ca07f2f5", "metadata": {}, "source": [ "Looks like the two AP-1 motifs are being pulled out neatly and the YY1 motif is a bit less clean. Hopefully, a seqlet caller will be able to identify all three patterns based on being much " ] }, { "cell_type": "markdown", "id": "4596ba00-e911-4ca2-a8d0-cf1fa0ab64d8", "metadata": {}, "source": [ "##### Recursive Seqlets\n", "\n", "A new seqlet calling algorithm implemented in tangermeme is the \"recursive seqlet\" caller. This method is meant to be a simplification -- both conceptually and implementation-wise -- on the seqlet calling algorithm used in TF-MoDISco. Conceptually, it works by first constructing a background distribution of attribution scores for all span lengths (between user-specified parameters) and then scoring all potential spans according to these background distributions. This process results in a `(n_lengths, sequence_length)` matrix of p-values. \n", "\n", "There are several ways that one could decode this matrix to get actual seqlets, many of which introduce several new parameters that would need to be optimized, but the recursive algorithm relies on a simple definition: a seqlet is the longest span whose p-value is below a threshold where *all* internal spans *also* are below the p-value threshold. This gives a precise meaning to what a seqlet actually is and also circumvents several problems that can arise in decoding with other definitions, e.g., the flanks can be too long when you say a seqlet is simply the longest span below a threshold if there's a central core that has really high attributions.\n", "\n", "This definition results in a method that has very few parameters and is quite fast. Let's see it in action." ] }, { "cell_type": "code", "execution_count": 3, "id": "5444e247-b9d9-4c0a-826c-9fae9cf9b139", "metadata": { "scrolled": true }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
example_idxstartendattributionp-value
1015781583-0.0240134.662937e-15
00103310390.4463364.798464e-03
20932937-0.0208145.285412e-03
309629670.3579376.717850e-03
4099810020.2788548.579600e-03
\n", "
" ], "text/plain": [ " example_idx start end attribution p-value\n", "1 0 1578 1583 -0.024013 4.662937e-15\n", "0 0 1033 1039 0.446336 4.798464e-03\n", "2 0 932 937 -0.020814 5.285412e-03\n", "3 0 962 967 0.357937 6.717850e-03\n", "4 0 998 1002 0.278854 8.579600e-03" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from tangermeme.seqlet import recursive_seqlets\n", "\n", "seqlets = recursive_seqlets(X_attr.sum(dim=1))\n", "seqlets" ] }, { "cell_type": "markdown", "id": "3e7d73b7-980d-4b47-a298-65a2b64168db", "metadata": {}, "source": [ "Note that the function is run on the the attributions corresponding to the observed characters and so is only two dimensions (with the dimensions being the examples and the other being the length of the examples). A dataframe is returned that is mostly BED formatted, with the first column being the index of the example and the second and third columns being the start and end of the seqlet call, respectively. The attributions are the sum of the attribution across the seqlet and the p-value is the value of the seqlet (at that position and at that length). The recursive property means that all internal spans must also have a p-value lower than the threshold, but those values do not affect the p-value of the seqlet returned here." ] }, { "cell_type": "code", "execution_count": 4, "id": "9d903c79-da28-46b9-aad5-1673ccd5155c", "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAA+kAAADtCAYAAADDRDT0AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAACDmUlEQVR4nO3dd3xT1fvA8U+6J6VQyii7QBVa9t5DQFC+LEEEARFBhnsgQxERLaC4QGSIIAjIntYfe4nsIXtvKN0FurN+f5wmbdq0tFBoCs/79eorzc3NvSc5N8l97nnOORqj0WhECCGEEEIIIYQQ+c4uvwsghBBCCCGEEEIIRYJ0IYQQQgghhBDCRkiQLoQQQgghhBBC2AgJ0oUQQgghhBBCCBshQboQQgghhBBCCGEjJEgXQgghhBBCCCFshATpQgghhBBCCCGEjZAgXQghhBBCCCGEsBESpAshhBBCCCGEEDZCgnQhhHjKrVy5koCAAPNfUFAQTZo0oW/fvsycOZOoqKj8LqIFU3mPHz+e5To3btwgICCAOXPmmJft27fP4nWm/3vnnXcYOXJklo+n/xs5cqTFvrZs2ULVqlWJjo4GIDQ0lHHjxtG+fXuqV69O/fr16dSpE59++imhoaHm502dOpWAgADz8zJ68cUX6du3r9XHoqOjCQwMzPZ9yPh6AgMDad++PT/99BPJyclZvne5derUKV599VXq1KlDQEAA8+bNy3LdrN5Dk2nTppnXuXHjRp6VMTdMx8m+ffse6X769u2bqX4DAgKYOnXqI9lfWFgYU6dO5fTp05keMx2LQgghbINDfhdACCGEbQgODqZixYrodDqioqI4dOgQs2fP5rfffuP777+ncePG+V3EPPHBBx/QoEEDi2WFCxfGzs6OXr16mZedPHmS8ePHZ1q/SJEiFs/duHEjdevWpUiRIty+fZuuXbtSqFAhBgwYQIUKFYiLi+PChQv8/fffXL9+nZIlSz70a1i7di1arRaA5cuXExQUZHU9FxcXfv/9dwDu3LnDX3/9xc8//8ylS5f44YcfHrocAKNHjyYxMZHvvvsOLy8v/Pz8sl3f3d2d//u//+PTTz/Fw8PDvNxoNLJq1So8PDyIi4vLk7I9iGrVqrFkyRIqVar02Pe9ZMkSSpQo8Ui2HR4ezrRp0/Dz8+PZZ5+1eKxHjx40a9bskexXCCFE7kmQLoQQAoDKlStbBHvt27fntddeo3fv3rz11lts3LgRHx+ffCxh3ihXrhw1a9a0+ljZsmXN/5tam7NbX6vVsnXrVt577z0Ali5dSkxMDMuWLaNMmTLm9Z577jmGDBmCwWDIk9ewYsUKihYtSqlSpfjrr78YNWoULi4umdazs7OzKHuLFi24efMmf//9N6NGjaJ48eIPXZbz58/To0cPWrRokaP127Rpw8aNGwkJCaFnz57m5Xv37uXGjRv07NmTpUuXPnS5TBITE3F1dc3x+h4eHlnW96OWX/stUaLEI7s4IIQQIvck3V0IIUSWSpUqxSeffEJ8fDx//vmnxWPHjx9nyJAh1K9fn6CgILp06UJISEimbURERDB27FiaN29OYGAgrVu3Ztq0aeh0OvM6pvT02bNn88svv9CyZUuCgoLo1q0be/bseeSv80Ht2bOHe/fu8dxzzwEQGxuLnZ0dRYsWtbq+nd3D/+z+999/nDt3js6dO9OzZ0/u3bvHhg0bcvz8GjVqAHDz5s1s1zt37hxDhw6lXr16BAUF0blzZ1atWmV+3NTtQKfTsXjxYnOa+v14enry3HPPsWLFCovly5cvp3bt2pQvXz7Tc3bv3s3QoUNp3rw5QUFBtG3blrFjx2bqKmBK2z558iTvvPMO9erVo23btgCkpKQwceJEmjRpQo0aNejTpw8nTpygdevWFun31tLdR44cSa1atbh69SqDBg2iVq1atGjRgokTJ5KSkmJRhmnTptGjRw/q169P7dq16dq1K8uWLcNoNN73vcmY7t66dessu12Yynf16lVGjRpFu3btqFGjBs2aNWPIkCGcPXvW4jW99NJLAIwaNcq8DdO+rKW7GwwGZs+ezfPPP09gYCCNGjVixIgR3L5922K9vn378uKLL3Ls2DF69+5NjRo1aNOmDbNmzcqzi1JCCPG0kZZ0IYQQ2WrRogX29vYcPHjQvGzv3r288cYb1KhRg3HjxuHp6UlISAjvv/8+SUlJdOvWDVABeo8ePbCzs2P48OGULVuWI0eO8Msvv3Dz5k2Cg4Mt9rVw4UJKlSrF6NGjMRgM/PrrrwwaNIgFCxZQq1atPHk9BoPB4gIBgIPDg/0cbty4kZo1a5pbpGvWrMnChQt5++23ee2116hVq5ZFSndOy5Od5cuXA9C9e3dKlCjB119/zfLly+ncuXOOnn/t2jUgc9p+epcuXaJXr14ULVqUMWPG4O3tzdq1axk5ciSRkZEMGjSIli1bsmTJEl5++WXat2/P66+/nuPX8NJLL/Haa69x8eJF/P39uXv3Lps2beLzzz8nNjbWaplr1apFjx498PT05ObNm8ydO5fevXuzbt06HB0dLdZ/++236dixI7169SIhIQFQwWlISAhvvPEGDRs25MKFC7z11ls5Tq3XarUMHTqUl156iddff50DBw4wffp0PDw8eOutt8zr3bx5k5dffplSpUoBcPToUSZMmEBYWJjFejkxbdo0i4sABoOBzz//nBs3bpi3Hx4eTuHChfnwww8pUqQId+7cYdWqVfTs2ZNVq1ZRsWJFqlWrRnBwMKNGjWLo0KG0bNkSINvW83HjxrFkyRJeffVVWrZsyc2bN/nxxx/Zv38/K1eutDh+IiIi+PjjjxkwYABvvfUWmzZtYsqUKfj6+tKlS5dcvWYhhBASpAshhLgPNzc3vL29CQ8PNy/74osvqFy5Mr///rs5wG3WrBkxMTF89913dOnSBTs7O6ZOnWruC20KKho1aoSLiwuTJk1i4MCBFn1/9Xo9c+fOxdnZGYCmTZvSpk0bfvrpJ+bOnZsnr+f999/PtGzjxo2UK1cuV9vR6/Vs3ryZN99807ysU6dOHDx4kGXLlvHPP/+g0WioWLEizZo1o2/fvpQuXTrTdpo0aZLlPurXr29xPzExkZCQEGrWrGl+355//nlWr17NtWvXLNL1TUwXAO7evcv69evZvHkzQUFBVlusTaZNm4ZWq2X+/PnmPvQtWrTg7t27/Pzzz/Tq1YsiRYqYAzUfH59cpWo3bNiQ0qVLs3z5cj755BPWrVuHvb09zz//fKaMDYBXXnnF/L/RaKRWrVrUr1+fVq1asXPnTtq0aWOxfpcuXXjnnXfM9y9cuMD69esZNGgQH374IaDedx8fHz744IMclVmr1fL222/ToUMHQB3HJ06cYP369RbBd/oLTwaDgfr162M0Gpk/fz7Dhw9Ho9HkaH8AVatWtbj/5ZdfcunSJaZPn27uTlGvXj3q1atnXkev19OiRQtefPFFlixZwqhRo/Dw8KBy5cqA6tJxv7q6ePEiS5YsoXfv3nz22WcW5enRowe///67xecoNjaW2bNnU716dQAaN27M/v37WbdunQTpQgjxACRIF0IIcV/pU3WvXr3KpUuX+OSTTwAsWoGbN2/Otm3buHz5Mv7+/mzfvp0GDRrg6+ubab1Jkyaxf/9+iyC9Xbt25gAdVP/gVq1asX79evR6Pfb29g/9Wj766CMaNmxosexBBnPbv38/MTEx5nRqAI1Gw/jx43nzzTfZsWMHJ06c4ODBg8ybN48lS5Ywa9asTIH3vHnzrLa2Wwse//77b+Li4ujevbt5Wffu3Vm1ahUrVqzIdAEiISGBatWqWZSvefPmfPnll9m+tr1799KoUaNM70vXrl3ZuXMnR44coXnz5tluIzsajYauXbuyePFiPvzwQ1asWEGHDh1wd3e3un5UVBQ//vgjO3bsIDw83CKN+uLFi5mC9Hbt2lnc379/P4A5wDZp3759jrMoNBoNrVu3tlgWEBDA3r17LZbt2bOHmTNncvz48Uyt9FFRUQ88rsOsWbP4448/mDBhgkX/f51Ox6+//sratWu5du2aeUBBUO/NgzCl0nft2tViefXq1fH392fPnj0Wx1qxYsXMAbpJQECA1ZHkhRBC3J8E6UIIIbKVkJBAbGwsVapUASAyMhKASZMmMWnSJKvPiYmJAVRQsm3bNotA0dp6JtYCGB8fH7RaLQkJCXh6ej7w6zApU6ZMlqOh58aGDRuoVq2a1dZxPz8/evfubb4fEhLChx9+yOTJk83p6iYBAQFWU8/TX6wwWb58Oc7OzjRr1oy7d++an+/n58eqVat45513LC5kuLi48McffwDg5OSEn5/ffdPvQbWMFitWLNNyX19f8+MPq3v37vz888/MmDGDkydP8umnn1pdz2Aw8PrrrxMeHs6wYcOoUqUKrq6uGI1GevbsaXU6OVM5078eyHx8OTg4ULhw4RyV19XVNVOdODk5Wez/2LFjDBw4kPr16/Pll19SokQJHB0d2bx5MzNmzCApKSlH+8pozZo1fPfddwwfPpwePXpYPDZx4kQWLlzIoEGDqFevHl5eXmg0Gj799NMHnmrP9H5lfB9Ny27dumWxzNp7mPG9EUIIkXMSpAshhMjW9u3b0ev15hZgb29vAN58802LVuT0KlSoYF43ICDAPPp5RhmDANMFgIzLHB0dcXNze9CXkOcMBgObN2/Och7zjDp27MisWbM4f/78A+/z8uXLHDp0CMDcpzijf/75x6KV1c7O7oEuSBQuXJiIiIhMy01dHkzHwMMoWbIkjRs35ueff6ZChQrUrl3b6nrnzp3jzJkzTJw40aJl9+rVqznelymIjIyMtBjRXqfT5ckFB5O//voLBwcHZs6caRHQb968+YG3uXv3bsaMGUPXrl0tUvhN1q5dS5cuXTJlXsTExFCoUKEH2qfp/QoPD8/Ubz08PDxP6l8IIUTWJEgXQgiRpVu3bjF58mQ8PT3Nc4hXrFiR8uXLc+bMmfv2523ZsiU7duygbNmyeHl53Xd/GzduZMSIEeYAJy4ujm3btlG3bt08SXXPK4cPHyYiIiJTWnV4eLjV1sf4+HhCQ0OtPpZTphb4CRMmZOp7npSUxPDhw1mxYkWOp0LLTqNGjdi0aRNhYWEWQe2aNWtwdXXNs6nCBgwYgLOzM88//3yW65j6cDs5OVkst9Z3PSumPtshISEWWR0bNmzI1aB996PRaLC3t7cYxT8pKYm1a9c+0PZOnz7N22+/TYMGDbLsoqDRaDINnLd9+3bCwsIsxlkwvX85ac03dQdZu3atRRr7sWPHuHjxIkOGDMn1axFCCJFzEqQLIYQA1HzXer0enU5HdHQ0Bw8eZOXKldjb2zNt2jSLlOwvvviCQYMGMXDgQLp27Urx4sW5c+cOFy9e5OTJk/z0008AvPPOO/z777/06tWLvn37UqFCBVJSUrhx4wY7d+7kiy++sGips7e3Z8CAAQwYMMA8BVRcXBxvv/12pvLu3bvX6jRieRGk3s+GDRuoUqWKOWPAZMaMGRw+fJiOHTvyzDPP4OLiwo0bN/jjjz+IjY1lxIgRD7Q/nU7HmjVr8Pf3z5TubNKqVSu2bt1KdHR0tiO358Tw4cPZtm0b/fr1Y/jw4Xh5ebFu3Tq2b9/Oxx9/nCfdDkANDNi0adNs16lYsSJly5ZlypQpGI1GvLy82LZtG7t3787xfipXrsyLL77I3Llzsbe3p2HDhpw/f565c+fi6emZq8HcstOiRQvmzp3Lhx9+yMsvv0xsbCxz5szJdIEhJ+Li4hg0aBAuLi4MHDiQEydOWDxeqVIlPDw8aNmypXkUd9P0c3PmzMnUAl62bFlcXFxYt24d/v7+uLm54evra3ERxqRixYq8/PLL/PHHH9jZ2dG8eXPz6O4lS5bktddey/XrEUIIkXMSpAshhADUFFUAjo6OFCpUCH9/fwYNGkSPHj0yBX0NGzZk2bJlzJgxg6+//pq7d+9SuHBh/P39LQbn8vX1Zfny5UyfPp05c+YQFhaGu7s7fn5+NGvWLFM6bp8+fUhOTmbChAlERUVRuXJlZs6cSZ06dTKV99tvv7X6OrZs2fKwb8V9bdq0yWLwNhPTNGh//fUXc+bM4d69e3h5eVGtWjVmzZr1wBcQtm/fTkREBIMGDcpynZ49e7Jx40bWrFnDgAEDHmg/JhUrVuTPP//ku+++Y/z48SQlJeHv709wcLB5er3HxdHRkRkzZvDVV18xduxYHBwcaNSoEfPmzcsy7d+a4OBgihUrxvLly5k3bx7PPvssP/zwA2+88cYDp4Vn1KhRI77++mtmz57NkCFDKF68OD179qRIkSKMGTMmV9uKjY01dzmwVp/z58+nQYMGjBkzBgcHB2bNmkVCQgJVq1Zl6tSp/Pjjjxbru7q68vXXXzNt2jQGDhyIVqvlrbfesnoBDNQUbGXKlGH58uUsWrQIDw8PmjVrxocffijp7kII8YhpjOmH7BVCCCHywY0bN2jTpg0jRoxg4MCB+V2cbB07dowePXqwdu1aAgIC8rs44iEcPnyYV155hW+//ZZOnTrld3GEEEIIQFrShRBCiFypXr06Z8+eze9iiFzavXs3R44cITAwEGdnZ86ePcusWbMoX758prEFhBBCiPwkQboQQgghnngeHh7s3r2b+fPnEx8fj7e3N82bN+eDDz6wOt2dEEIIkV8k3V0IIYQQQgghhLARdvdfRQghhBBCCCGEEI+DBOlCCCGEEEIIIYSNkCBdCCGEEEIIIYSwERKkCyGEEEIIIYQQNuKpDtINBgPHjh3DYDDkd1FENqSeCg6pq4JB6qlgkHoqOKSuCgapp4JB6qlgkHp6tJ7qIN1oNKLVapEB7m2b1FPBIXVVMEg9FQxSTwWH1FXBIPVUMEg9FQxST4/WUx2kCyGEEEIIIYQQtkSCdCGEEEIIIYQQwkZIkC6EEEIIIYQQQtgICdKFEEIIIYQQQggbIUG6EEIIIYQQQghhIyRIF0IIIYQQQgghbIQE6UIIIYQQQgghhI2QIF0IIYQQQgghhLAREqQLIYQQQgghhBA2QoJ0IYQQQgghhBDCRkiQLoQQQgghhBBC2AgJ0oUQQgghhBBCCBshQboQQgghhBBCCGEjJEgXQgghhBBCCCFshATpQgghhBBCCCGEjZAgXQghhBBCCCGEsBESpAshhBBCCCGEEDZCgnQhhBBCCCGEEMJGSJAuhBBCCCGEEELYCAnShRBCCCGEEEIIGyFBuhBCCCGEEEIIYSMkSBdCCCGEEEIIIWxEvgfpCxcupHXr1gQFBdGtWzcOHjyY5bobN25kwIABNGzYkNq1a/Pyyy+za9eux1haIYQQQgghhBDi0cnXID0kJITg4GCGDh3K6tWrqVOnDoMGDeLWrVtW1z9w4ACNGzdm1qxZrFy5kgYNGjB06FBOnTr1mEsuhBBCCCGEEELkvXwN0ufOnUv37t3p0aMH/v7+jBkzhhIlSrB48WKr648ZM4ZBgwZRvXp1ypcvzwcffEC5cuXYunXrYy65EEIIIYQQQgiR9xzya8cpKSmcPHmSwYMHWyxv0qQJR44cydE2DAYD8fHxFC5cOFf7NhgMGI1G9Ho9gPlW2Capp4JD6qpgkHoqGKSeCg6pq4JB6qlgkHoqGKSeHoy9vX2O1su3ID0mJga9Xk/RokUtlvv4+BAREZGjbfz2228kJibSoUOHXO37xIkTaLVa8/3jx4/n6vkif0g9FRxSVwWD1FPBIPVUcEhdFQxSTwWD1FPBIPWUO3Xq1MnRevkWpJtoNBqL+0ajMdMya9avX8+0adOYPn16pkD/fgIDA80t6cePHycoKCjHVzXE4yf1VHBIXRUMUk8Fg9RTwSF1VTBIPRUMUk8Fg9TTo5VvQbq3tzf29vZERkZaLI+KisLHxyfb54aEhDBmzBh+/PFHGjdunOt929lZdsW3t7eXg6sAkHoqOKSuCgapp4JB6qngkLoqGKSeCgapp4JB6unRyLeB45ycnKhWrRq7d++2WP7vv/9Sq1atLJ+3fv16Ro4cyZQpU2jZsuUjLqUQQgghhBBCCPH45Gu6+4ABAxgxYgSBgYHUqlWLJUuWEBoaSq9evQCYMmUKYWFhTJ48GVAB+ieffMLo0aOpUaOGue+6i4sLnp6e+fY6hBBCCCGEEEKIvJCvQXrHjh2JiYlh+vTphIeHU6VKFWbNmoWfnx8AERERhIaGmtdfsmQJOp2O8ePHM378ePPyrl27MnHixMdefiGEEEIIIYQQIi/l+8Bxffr0oU+fPlYfyxh4L1iw4HEUSQghhBBCCCGEyBf51iddCCGEEEIIIYQQliRIF0IIIYQQQgghbIQE6UIIIYQQQgghhI2QIF0IIYQQQgghhLAREqQLIYQQQgghhBA2QoJ0IYQQQgghhBDCRkiQLoQQQgghhBBC2AgJ0oUQQgghhBBCCBvhkNsnREZGMmnSJPbs2UN0dDRGo9Hi8dOnT+dZ4YQQQgghhBBCiKdJroP0kSNHEhoayrBhw/D19X0UZRJCCCGEEEIIIZ5KuQ7SDx06xKJFi3j22WcfRXmEEEIIIYQQQoinVq77pJcsWTJTirsQQgghhBBCCCEeXq6D9NGjRzNlyhRu3LjxKMojhBBCCCGEEEI8tXKd7v7++++TmJhI27ZtcXFxwdHR0eLx/fv351nhhBBCCCGEEEKIp0mug/TRo0c/inIIIYQQQgghhBBPvVwH6V27dn0U5RBCCCGEEEIIIZ56uQ7SAfR6PZs3b+bixYtoNBoqVapE69atsbe3z+vyCSGEEEIIIYQQT41cB+lXr15l8ODBhIWFUaFCBYxGIzNnzqREiRLMmjWLsmXLPopyCiGEEEIIIYQQT7xcj+4+YcIEypQpw/bt21m1ahWrV69m27ZtlC5dmgkTJjyKMgohhBBCCCGEEE+FXAfpBw4c4OOPP6Zw4cLmZd7e3nz00UccOHAgL8smhBBCCCGEEEI8VXIdpDs5OREfH59peXx8fKbp2IQQQgghhBBCCJFzuQ7SW7ZsydixY/nvv/8wGo0YjUaOHj3KuHHjaN269aMooxBCCCGEEEII8VTI9cBxn376KZ988gkvv/wyDg7q6Xq9ntatWzNmzJg8L6AQQgghhBBCCPG0yHWQXqhQIX755ReuXLnCpUuXMBqNVKpUiXLlyj2K8gkhhBBCCCGEEE+NB5onHaB8+fKUL18+D4sihBBCCCGEEEI83XIUpAcHB/Puu+/i5uZGcHBwtuuOGjUqTwomhBBCCCGEEEI8bXIUpJ86dQqdTmf+XwghhBBCCCGEEHkvR0H6ggULrP4vhBBCCCGEEEKIvJPrKdhGjRpFXFxcpuUJCQmS6i6EEEIIIYQQQjyEXAfpq1evJjk5OdPypKQk1qxZkyeFEkIIIYQQQgghnkY5DtLj4uK4d+8eRqOR+Ph44uLizH937txh586dFClS5FGWVQghhBBC5BOjET76CK5fz++SCCHEky3HU7DVrVsXjUaDRqOhffv2mR7XaDS8/fbbeVo4IYQQQghhG3bsgClTQKuFH3/M79IIIcSTK8dB+vz58zEajfTv35+pU6fi5eVlfszR0ZFSpUpRvHjxR1JIIYQQQgiRvzZuVLfr1kmQLoQQj1KOg/T69esDsGXLFkqVKoVGo3lkhRJCCCGEELbl4EF1e/mySnkvUyZ/yyOEEE+qHAfpJrdu3eLWrVtZPl6vXr2HKpAQQgghhLAtRiMcOpR2/+hRCdKFEOJRyXWQ3rdv30zL0reqnz59+uFKJIQQQgghbMqVKxAdnXb/+HHo1CnfiiOEEE+0XAfpBw4csLiv1Wo5ffo0P/74I++//36eFUwIIYQQQtiG8+ct7ycl5U85hBDiaZDrIN3T0zPTsiZNmuDk5ERwcDArV67Mk4IJIYQQQgjbcPNmfpdACCGeHjmeJ/1+ihQpwuXLl/Nqc0IIIYQQwkbcuJHfJRBCiKdHrlvSz5w5k2lZeHg4s2fPJiAgIE8KJYQQQgghbIe0pAshxOOT6yC9S5cuaDQajEajxfKaNWvy1Vdf5VnBhBBCCCGEbQgNze8SCCHE0yPXQfqWLVss7tvZ2VGkSBGcnZ3zrFBCCCGEEMJ2pB/ZXQghxKOV6yDdz8/vUZRDCCGEEELYKAnShRDi8cl1kA6wZ88e5s2bx8WLF9FoNFSsWJH+/fvTuHHjvC6fEEIIIYTIZ6YgvWRJCAvL37IIIcSTLteju//xxx+88cYbuLu7069fP/r27YuHhweDBw/mjz/+eBRlFEIIIYQQ+cRoTAvSp06F117L1+IIIcQTL9ct6TNnzmTUqFG8+uqrFstr167NL7/8kmn5/SxcuJA5c+YQERFB5cqVGT16NHXr1rW6bnh4OJMmTeLEiRNcvXqVvn37MmbMmNy+BCGEEEIIkUOJiZCSov6vVw+cnODAgfwtkxBCPMly3ZIeFxdHs2bNMi1v0qQJ8fHxudpWSEgIwcHBDB06lNWrV1OnTh0GDRrErVu3rK6fkpKCt7c3Q4cO5Zlnnslt0YUQQgghRC6ZTu+KFYMyZcCiLcVohCMfQfSRfCmbEEI8iXIdpLdu3ZpNmzZlWr5lyxZatWqVq23NnTuX7t2706NHD/z9/RkzZgwlSpRg8eLFVtcvXbo0n376KV26dMHT0zO3RRdCCCGEELmUlKRuq1UDjUb1Sy9RIvXBiJ1wZgoceivfyieEEE+aHKW7z58/3/y/v78/M2bMYP/+/dSsWROA//77j8OHDzNgwIAc7zglJYWTJ08yePBgi+VNmjThyJFHezXWYDBgNBrR6/UA5lthm6SeCg6pq4JB6qlgkHoqOJ70ukpMBHd3KFcOTC+xcuXU/0N3gJ07RP8HsRfAs0K+ljU7T3o9PSmkngoGqacHY29vn6P1NEaj0Xi/lVq3bp2zjWk0meZRz0pYWBjNmzdn8eLF1K5d27x8xowZrFq1ig0bNmT7/L59+/LMM888UJ/0Y8eOodVqc/08IYQQQgghhBDiQdSpUydH6+WoJX3r1q0PVZjsaDQai/tGozHTsrwWGBhobkk/fvw4QUFBOb6qIR4/qafHwMsr87I7d3K9GamrgkHqqWCQeio4nvS6OnwYWrWCMWNgxAi1TKcDBzsdrPYDQ2o+fIV+UGdq/hX0Pp70enpSSD0VDFJPj9YDzZOeF7y9vbG3tycyMtJieVRUFD4+Po9033Z2ll3x7e3t5eAqAKSeHiFrgz4+xHstdVUwSD0VDFJPBceTWleJiepnwtU17afB3h6Iuw66qLQV9fEP9dvxuDyp9fSkkXoqGKSeHo0cBenBwcG8++67uLm5ERwcnO26o0aNytGOnZycqFatGrt376Zt27bm5f/++y9t2rTJ0TaEEEKIRyU6MZrw+HCqFK2CnSbX46wK8cQwDRzn4ZHhgYSrj70sQgjxNMhRkH7q1Cl0Op35/6zkNk19wIABjBgxgsDAQGrVqsWSJUsIDQ2lV69eAEyZMoWwsDAmT55sfs7p06cBiI+PJzo6mtOnT+Po6EilSpVytW8hhBAiOzMPzmT01tHEfBJDYZfC+V0cIfJNlkF63JXHXRQhhHgq5ChIX7BggdX/H1bHjh2JiYlh+vTphIeHU6VKFWbNmoWfnx8AERERhIaGWjynS5cu5v9PnjzJ+vXr8fPze6T95oUQeW/PHihVSo0WLIQtStIlWdwK8bTKMkhPvPXYyyKEEE+DXPVJ1+l0VK9endWrV1OlSpU8KUCfPn3o06eP1ccmTpyYadnZs2fzZL9CiEfMWmZN6mQS167Bc89BlSpqQKJHPFakEA8kUZeobrWJ+VwSIfJXlkG6NvZxF0UIIZ4Kuepk5+DgQKlSpTAYDI+qPEKIp8DKlZCQAEePwr//5ndphLDOFJybgnUhnlZZBukpserWwR2Qq61CCJFXcj0SztChQ5kyZQqxsbGPoDhCiKfBunVp/69fn3/lECI70pIuhGIK0t3cMjyQEqNua/8Az454nEUSQognWq6nYFuwYAFXr16lWbNmlCpVCrcM39irVq3Ks8IJIZ48SUmwc2fa/f/+y7+yCJEdc5AuLeniKZeY+hFwdMzwgCndvfhzUKwZnPz6cRZLCCGeWLkO0tu0aZPrUdyFEMLk3DlInSwCMHdTF8LmmAaMk5Z08bRLTla3maZCTokFt9LgUV59mbuVfswlE0KIJ1Oug/S33377UZRDCJGfTJGyRvPIo+bUWRSFsHmm4DwvR3e/efcmU/dPpW/1vlTzrZZn2xXiUTINRZQpSNfGgoe/+l+jgaL1H2exhBDiiZXrPult2rQhJiYm0/K7d+/Spk2bPCmUEOLJdf58fpdAiJx5FOnu1+5cY9LuSZwIP5Fn2xTiUTNdu83ckh4Drulaz71rPbYyCSHEkyzXQfrNmzetju6ekpJCWFhYnhRKCPHkunEjv0sgRM6YR3fPw3T3eG28xa0QBYHVlnSjUaW7OxdJW+ZW5nEWSwghnlg5TnffsmWL+f9du3bh6elpvm8wGNizZw9+fn55WzohxBPn1i11W6KElUGIhLAhOWpJtzZGSzZdRhK0CRa3QhQEVoN0fRIYdeBYKG2ZjFkkhBB5IsdB+vDhw83/jxw50nIjDg74+fllWi6EeDKtPbuWkPMh/PLCL7keSNIUpI8aBTVqwMSJj6CAQuRWchRc+g2qvA32LsCjGThOgnRREFlNdzemjgDq6PXYyyOEEE+6HAfpZ86cAaB169YsX76cIkWK3OcZQogn1c6rO5l5aCY/PP8DLg4u1ldKPxhduvtRUepup05Qvjz88cejLasQOXLoXbi6EBJvQe3vARWcO9s75+nAcfEp8Ra3QhQE1tPdTUF6oUzrCyGEeDi57pP+zjvv4OHhkWl5SkoKq1evzosyCSFsXFxKnMVtrp4bB8WLQ4UKKn5v3Votj0mMYePFjUQnRudlUYW4P1083Fil/r8wA7R3AZXm7u3qnacDx0lLuiiIrAbpBq26dZAgXQgh8lqug/RRo0Zx7969TMvj4+MZNWpUnhRKCGHbzINfPUBrYFwcVKyYdr9ePXV74NYB2v/Rnn039uVFEYXIudANoE8NmvVJEL4LUC3phV0KS7q7eOqZEqPs0p81Gkwt6Z6Z1hdCCPFwch2kG41Gq31Qw8LCLAaTE0I8ucwpu7kcoVqng6QkKFkybZl/6hS7D9M6L8RDiT5ked+oxWg0kqhLDdLzsCVdRncXBZHptM9ich9Turud02MvjxBCPOly3Ce9S5cuaDQaNBoN/fv3x8Eh7al6vZ4bN27QrFmzR1JIIYRtedCAOj41Lkk/pIXp5E+CdJFvYo9lWqQz6DAYDdKSLgRpLeh6fbqFpiBdk3HydCGEEA8rx0H6c889B8Dp06dp2rQp7u7u5sccHR3x8/OjbNmyeV9CIYTNedB097jU+NvauJMSpIt8E3810yJT67m3izdJ+rwbOE6CdFFQrFkD33wD8+ZlEaSb0t3tcnwqKYQQIody/M361ltvAeDn50fHjh1xdnYG4N69e6xdu5Y5c+Zw5swZTp8+/WhKKoSwGQ+a7m4K0r29rTwmQbrIL0m31a1nZbh3Hkibdq2wS2Fux93O+rlZzGKQlXhtPB5OHpLuLmxaSgq89x5cuQLDh0Pdumq5tKQLIcTjkes+6V27dsXZ2Zk9e/bw0Ucf0bRpU/744w+aN2/OihUrHkUZhRA25kEDagnShc0x6CA5Epy8ocMxKNMDSGtJz+s+6QnaBIq6FpWWdGHTFixQATrAxo1gGi/Yaku6xgGiDsI/PeDwh4+zmEII8cTKVY7S7du3WblyJStWrCAxMZEOHTqg0+mYOnUqlSpVelRlFELkp+RICNsGpbua0xofNt3dyiyOEqSL/JEcARhVcG7vAgHvQHKkeW50bxfvPO+TXtRNgnRh29assbyflNrjI8uW9KRQuL4cvAIfS/mEEOJJl+OW9EGDBtGxY0cuXLjAZ599xq5du/jss88eZdmEELbg4Nuwuyccfs+8KD4lHk8nz1yn7CakxiUOVi4PSpAu8kViaip7kdR83qKNwNmHRG0iGjQUci6Ut6O7p8RT1LXoA01fKMRD0Wgy/1lhNMK//2Z+KmQVpEufdCGEyGs5DtJ3795Njx49eOedd2jZsiX29tIHSYgnXuxxuPan+v/CLxB3BaPRSFxKHMXci+U6oDZN3+PomPkxc5CulSBdPCLWghRTf3T38urWzh6KNiBRl4iLgwuujq7mVvUHcegQnDmTdl9a0oWtCw2FqCjLZdkOHKfJdc9JIYQQ95Hjb9aFCxcSHx9P9+7d6dGjB3/88QfR0dGPsmxCiPx2dXHa/0YDxF0iSZeEESPF3IrlujXQdIKXVUt6Udei0pIuHq+U1N8xl2Jpy+wcSdSqIN3FweWB091PnYKGDaFOnbRAPUGbgI+rjwTpwmadO6du7exg/HhwcwOn1KnQLYJ004BxRgNCCCHyVo6D9Fq1ajFhwgT++ecfXn75Zf766y+aN2+OwWBg9+7dxMXJibUQT5yI3ZkWmVLci7kXy3W6+/1a0n3dfSVIF4+XIUXdOhezWGxqSXdxcHngdPd33gGdTnXzGD9eLYvXxlPUrSjx2niM9xkJXoj8cOmSuu3SBT77DD75BFIn9EGrTbeiXeoXuSntPQOjMa2LkxBCiNzJdY6Sq6srL730EosXL2bt2rUMGDCA2bNn07hxY4YMGfIoyiiEyC/3zmZaZAqii7nlPt39fi3pEqSLx84cpPtYLE7SJeHq6PrALemXL8OWLWn3r6ZOxW4a3d1gNJCiT3nQUgthlqxLJjoxGr1Bf/+Vc8CUJNmtm7odMCCtJd0i6Db1RTda3++AAeDlBcuX50mxhBDiqfJQHYkqVqzIiBEj2LFjB999911elUkIYQv0SZAUBvau0Gw1uJcD0kZ093HzeeCWdAnShc0wpKhR3e2dLRZbpLs/QEt6+gA9PVOfdNP/QjysRccXUXRyUU6En7BYHhqqplKLz+5r2so4DaYgPShI3ZYpA+XU1z8WSZN2WQfp69fD77+rTJIhQzI8TwghxH3lyWgf9vb2PPfcc8yYMSMvNieEsAUJ19VtxQFQujPUmgKodF0neycKORfK1Cd91y51a5pTNyOrIwSnikuJo7h7cQnSxeOlTwFN5v4X5oHjHFxJ1CbmOjX98GF1a2eX1gppNBrNo7sDub7IJYQ1MUkxFregguOuXaFfP+jeXaWek/4YNhot76ffXupmSpdOW+blpW4tgm1zS3rmdPcpU9L+j4qC27dz+mru4/J82NIaog/l0QaFEMI2yZCcQgjr4lPzc4s/p279OoOLaul2d3TH3dHdIqC+eBFefln9/+ab1jdpmhRCZ6ULY1xKHMU9JEgXj1D6wMT0vyElrW9tOulb0o0Y0Rq0mdbJzonURs3vv4ejR1Wf3mR9MkaM+Lip1HppSRd5ITox2uIWYNo02LdP/b9hAxw8mIvtRavj1ds7bZmLi7q12pKeIc3+3r20C7Z56t5FODAEwrfBzk6gzeJqsBBCPAEkSBdCWGdqSfespG7tHMCrGvEp8bg5uuHm6GbREvjBB2lplX/9BZGRmTdpmsYnqyDdlO4uA2qJxyZ9kH55ARwYBjfXkahLxNVB9UkHLPqlL1oEvXrB2cxDNpjdvKlGxR48GJ59VrVomoLyIq5FAAnSRd6wFqT/8YflOom56LEREwMlS1pOo241SE/fkl6iHVT7FFAXpUzZUtWq5Xy/93X2e9CnvpDEULh7Jvv1hRCiALPSM1QIIUhrpXD1S1um0RCvjcfdyR13J3dzuvvdu/D332lpvZDW/zw9U0u6NkOjZIo+Ba1Bi6+7LwajwTxolxCPXPogPWwbXJ4Lzj4k6ezMLemg0t+98GL3bujfX11oOnAATp5MC2DSi4iA2rXTHuveHWJTPy/uTu64OLjkegpDIazJGKTHxsKhh8gGj4kBT0/LZa6pX8dWg3SDVo3p4OAOwLFjanGXLrBypRpALk+EbsijDQkhhO2TlnQhhHX6JDWglpO3xeK4lDhzS7opNX3fvsyBtzVZpbubtuPr7mtxX4hHzqBNCzbSSZ/ubroPajoq0/F76ZJqMc8oOVml/FaokLbM0zOt5dzVwRVXB1dpSRd5wtwnPVHdmsZD8PFRae4tW2bzZCtdQJKTLS+4QtrFpvTjjWj1qZ8bneX39a1b6nboUNUa//nnlq3yD0SXCHEXAQ34/Q80cvoqhHiyybecEMI6QzI4emU6u4pPiTf3STelu5tabUx90l980fomTSd+GQN6CdJFvrFzAGPmtI9EXSI3791k9ZnV5vu3b8Pu3fffZESEujWNiG1iDtIdXXFzdJMgXeSJ6MRoCrsUNrekX76sln/0EdSpA7NmpXU1ygm9Pm1edBNTkB6TNjYdN26lBunauxbrhoer29q11W2FCpYXrHLqv//UCPGJiUDcBcAIFV+H5mug+le536AQQhQgEqQLIazTJ4Gdc6bF8dp0fdJT03Vv3FCPDRumbsePt95y4uGhbpOSLJdLkC7yjZ0TGDOngSRqE9l7Yy+D1w8G1Lzp//yjHitWTAU+pUpZ36RpPIYSJSyXJ2gTcLJ3wsHOQYJ0kWeiE6OpULgC0UkqSA8NVctNLeiVK0Pdujnfnk6XdUt6+lHaL17OOkj38ICiRdOW5eYiAcD+/dCkCbz2GnTsCPrY1AEgyr+qbqu8C87FcrdRIYQoQKRPuhDCOn2SCmAAriyE2ONQqqMa3T21T3qiLhG9QU9oqD0ODhAYqEa19ve3flJmCtJjYy2Xx6XEYaexw9PJE0c7RwnSxeNj56RS3jPIODd6ojaR06fV/19/DW+8oeaPtsaUEuzmZrk8XhuPq4Pq3Ovq6CpTsIk8EZMYQ80SNc3p7qZAulKltHWsjZuQFb0eHDKcHZqeb7oAAHD+giPPFQV0lkF6WBiUL//gKe5GI7z+etpApNu3Q0LEFTwBvFJHonNwBY/yD7YDIYQoAKQlXQhhnSFJDQYEcGM1nJ4EUfst0t1BtQ6Ghqp0xvQndtm1pEdHWy43Teum0Whwd3KXIF08PnaOVoP0JJ1lukeiLtHc/7xDB3Xbvj34+ZGJqc96xpThBG0CKfoUPtzwIWFxYdKSLh6a3qAnNilWtaSnprvfvq3mNS9S5MG2aW+fuUuSKUg/cUIF0bduQWi4ixrPIUNLelQUFC78YPsGOHJEDciYnp0+Dhw8wNnH+pOsdFkRQoiCTIJ0IYR1RgOQOdKO18ZjMBq4m3zXfD80FMqWvf8msw3SnVTQn3H+dSEeqfQt6bUmQ+EagPWW9Js3oVChtDR3jcZ6C6UpwMn4WII2gURdIt/t/Y6weAnS72ftWqhfXw1AliBvlVV3ku9gxGgRpEdEqDnOH7Ql28EBUlIsl7mrr2ciI+HaNTWzAWjAqTAkW36hp6TkruU+o61b1W2XLqpV/vnnAX0CuBS3/qJOfwNLnGFTY0iwMpKjEEIUQBKkCyGss3ex2sIYlxLH4hOLqf9rffP9u3fTTuKyYwrS0w8+ZNqGBg3br2w33xfisbBzUnMvG3SqlS41eyT9vOiQ1pLu73//4Mc0WHbGLh8Zp1yTKdiytmkTdO2qgsEZM+CHH/K7RLbJFJhX8E4L0lNS0qZMexAODmqGgvRcXdMyQ/bvh1WrUh9wLAyJloGxVvtwQfr58+p20iTw9YXffgMHTSLYW3lRoRvg6Ag1V3vkHjg//cF3LIQQNkSCdCGEdXYuaoT3DDL2o41PiUerzdlJoZOTOgE0jf4LqtUnLiWO0LhQWv3eipv3bkqQLh4fOyfACCmWV44ytqQn6ZKIiVEt6fdj6vaRMdDJ2HKe05b0u3fvv86TZsIEMKTLYM442KRQYhJjsNfYU7pQae6l3EOr1+b4+zgrzs6ZW9I1mrT0+eHD1ajrgGpJj7tksa7BkLlPe25cuaIyASpXVvdLlgRn+wR14RhgT39YWwGuLIJL89Qy15KpA50as9/49ZXqNvbEgxcwB+7dgw8/hIED4dSpR7orIcQTSoJ0IZ50Ufvh1CSIOpi759k7gyH1TM2nMTh4AlZaA7Xx6HSZ+99ao9Go1vQrV9KWHTwI95LvWax3vyD9xAno319NMWQaSVuITAx6CN8Bd89kvY5pcMRkywMpU590bSIpKZkHg7PGFKBkDCxzE6Tr9TB6tPq/YkU1mvzTIiYGdu2C0qVVavWOHTnL1Cmo4lPiuRB9IdMxlxPRidF4u3rj7eINQGxSLEZj7kdTT69IkcyDe4IKnCFtikFAtaTHX4bEtBHlrLXEZ0sXr8Y9SQ2cr16FatUyZKzok9M+q0lhEH9F9YWP2AWFq8OL56HjcXD2zXo/J4Nh3wD1/5aWkHAjF4XMuaQkaNwYvvtOZQG0a/dIdiOEeMJJkC6eLGHb4Mz3cEcuXQNwbanqp/ffSNhYD8J3qeWxJ+DAMDj5NaTcsf5cexd18gTwzPvgVhrIHEDHpcRZHWgoK6Yg/U7qbnftsr7NrOzdq6YTmj8fpkyBd9/N2X7FU0aXCNufVyfjfz0LJyZYX8+psLrNEKRbS3e3Nn+0NaZU34yBSqYslGxGd58zB37+Wf2v1cLKlfff75Pi9GnVZWDYMDWCfvPmadM7PomWnFxC5amV2Xp5a66fa5ojvbBLYfN9R8eHyzzw9laDzxkyjMVmdSA60+cnbBvcUdMf5Gr/ibdhQ33Y1RX+DoJzPxMRYWVf6S8am2hjVap9hQHg4A6elaFyFgeKLgFOfpl236gF3aPJ2Pr9d3UhuVAhqFVLxlMQQjwYCdJFgWU0ZkjJO/k1bG0NRz6Av6tn33r2NDAa4ehIMOqhWDM1Mm5KFEQdgA114cIvcGwMHPnQ+vPtXEB7RwU76VhLd3d0zPmJiJeXOvn7v/9TLSZHjmQRpGs0mf+ATz9Vwc9zz8GLL+b84oB4AEYDhG6C21usjk9g0y7MhLDN6uTd3gViDltfz6W4uk24brHY2sBx9vY5ayH0SR2AOirKcnluWtKnT1ejbAMsWWJ9FPkn1dnUKbGbNElb5umZP2V5HM5GnrW4zY3oxGi8XbzxcPLAXmNPdGI0Dg6QmHj/52alSBE1Q0HGLKVsg/QDb8KVBYBKtc/x/k9PgrunVAu4vRvcPUNyctr4JWb2bmpa0PS0qRlYReqkWy/DBO8m4TvV2BM1J6v79X/jUZ0Cr1unPrs7d8Lhw+pispnRmDZohRBCZEOCdFEgLV6sTlpdXKBjR0i8exdOfKF+6P0HgWuptB9wG3Qv+R4Xoi+Qok+5/8oP6s4JlYZYYxI8txM6ngDX0nD2e9UiEfA+VBqS9dQ1qfM5k3TbYrG1dHcPj5wH6aaRsadNS2sdy2lLemKimjO3RQvYuFGdDH3ySc72K3JJl6Baobe3g23PqVbpguTGCvCoCJ0uQZdbUOI56+uZgvTY4xaLE7WJfN7icza+upE6JeuQqEvEySlnx3mxYur22jXL5TkN0hMS4Phx6NRJ3X/+efjxx/vv96HEXYHIfZkDoXxget/Klcvfcjwu56LPYa+x51zUuVw/NyYphrNRZ2k2txlGjMQkxeDtDXEP0UhsCsavp163MhpVFwTTd7cFx8LqVheHqT+4r68alf2+jEbVR9ynEfzvMnS+Cj6NrKfrO7ipEd4Ban8PzsVU0A3gWiI1+DVk/XsWuVuly5fvp+6X7a5a3h+BY8egYUOooSaKYEBqhj031sDasrDUBXb3KngXPoUQj5UE6aLA2bMH+vRRgyk1aaKma0m5sl4Fnk2XQv1Z0OE/c3q2LZq2fxqVp1Y2j2b+SEQdULcV+qpb93LgXRNuroeKA6D2d1DvF6ieRRqwa2rTXYJlpGEtoC5RAvMc0vdjahH85x8ICUndhjbDNrVxmVscjEZOnFB9dXv0SOuvWKcO4iElJcEXX6iBmmrUgIULgUtzVX9Pv04Q8N4jSw19IPqU7C/CGfQQ+S+U7wsuvuDknXUarClIv7FCHW961VSeqEukgV8D2vq3xa+QH0m6JLy9czaIm6enGiQxfZBuMKgLWv7e/vSs1pPqxatnme5+4oRav3nztGWZWhbz0tFRsN4fNjWENeUyDaL3uJlaYXMySN+T4GzkWRqVacTZKMuW9MOHVer0oUNZPzc6MZq7yXfZfX03BqOB6MRoSpRQQfKDtqab+p4fPapuz5xRXZSsZnOYWtLTKV5cHfs6XYYHYo/Dtnaw2g92doHow+r3peJAFYQ7+0D53jg5QXzGj4a9q0qNNxrB61mV/m4K0h08VebMn/awvaP1F5UcCYWqpl18hgefoy4bcXHq4kam36X4a7C7B6TEqmkery+ziQtiVmWRxSaEeLwkSBcFzm+/qdtdu9Tff/+BS/IpdWW9WOpZrZO3Gu3VRh0KVWddh25lc/ZlcuckXFmsgm5T0JoUoZbd/Eu1eFqTHAmOXup90CeBNk6l9OruQakX0tZzs9Y8ArilTnxuCvZTxWvj+ajRR/z2v9/w9/YnPiWekiXh4sXMfRitsdYaE5cSh6+7L7VL1qa4e/EsW9IvXFC3AQGZH4uMhF9+gcmT1WB0IufefRfGjVNBUdGi8McfwNXFUKQeNFutWq6ar3+0hYj5Dy7+Bre3qiDbGqMRjn8BK4vA8kKqL6veSv55SrRqUUud8zxb9i7gWAjunYd/XoLY/wA1cJyLg+pc7uLgQqI2ET8/dQzeL1tVo1Gt6ceOpa27e7dqOe9UpRNLXlrCgJoDVEu6lRPi26nJKyUfx1dY1EE4PRGKPwd1fwavqpm6uDwwo0EFJVm1bmbBFBM8DVnBOoOOC9EXeN7/eXOQbjCoC9F16sBrr6kxODZtsv5807Rr6e+XKKH+v5Ru0PXcvJem7hqLFqnbmTPVbWlr171dM3+h+/qqi6lXr6r7BgPcjU1R2ThR+6BofYg5BNGpX9SFq1s8v1AhNTq6BQdP1ZKenG7UOk3qKawxi++L9FKi1XmByb43IOJfdcHv4m/q/vEvHnqeddNYK5m6Blyerz4H7fZD+/3Qdh/YOT7UvvKC0QhbtkBwMCxYYOXiiBAi3zy9Qfq1ZbCxkfp/S0t1giYKhF27oEEDNSALqIDN2RgBbmXU2d2VxXDyK3WV3kYdvHWQuqXqcjA0LZqMS4nj611fs+3ytrQVD70HIYGw/w3YWB+uLYFbISplbt9r8E932Pua9Z1oY1XwAWqQuOWecCpY3XctpQaPO/SuGvndGvfUIP3CTHXSro1FbzCQpEui67NdGVBrAOUKlyNeq4L05OS0IPr27cxzoZtUrJh5WVxKHG/WeZNDgw8xrN6wLIN002BEGfun/vsvlC+vpryZM8eyL2tB8d2e76g7qy7X7ly7/8p5KD4e5s5VKdUHDqjMlBkzgLgLUKpj2omwS7FHV4gjH8P/1VTjI2xvBxdnAzB592ScJziz6WJqhBK6AU6MgzI9oOHv6ji2ljKaktoZ3KkwRPwDywtDSPXM65m4pEY1N9TobDqDAZ1BZxmk61SQfu8e3LqlVjcasx4gy9cXbtxQ31fJyeo9TdAm4Oaohod3c3TLMt3dNM7Cw8w1nWNXF6ngpfkalW3QeovKPnjo7S6BteVghTes9IWw7dbXs3KRwjSC/iObes6Gov+rsVfRGrS0r9SeW/duEZcSx9q1KkAePFi1YK9bB4ULq8B340b4/HPVz/niRZXunl5MYow5SDddrLx7V439kVOVKqnbrVuhXz+YOlXdt9qS7pa5T0Lx4mnPBzX+SPy5NZB4S13sa7YKXjgHzkXVCo5e6nhZ6gpbWlGuHJzLmPnvkfrDcedk2jLTvOnau1AoAAo9k/WLMurTvstAtWTHX1WjvR8YpM4Bry4096tXz8lB//GkcDj+uRr47tA7EHcZsNL4HHMEijVVWQAAReuqC4QpMWqw2/2D1bg6iTnpJ5A39Hro0kWN7zJrlrpY+8svWL7udP/rDDpik2LR6iVNX4jH4ekM0u+egX9fSWtlKd1NfcmLAuHmTRWQWTCkgF3qvEeX58GxTyH6ALYoMiGSq3eu8lqN1yxa0rdc2sKYrWOYsCs1/TzmGJz7UQ100yMO/ncNvILUiYBPU+geCy/FQkAWw5s7uKelA2ZsmtLYqf7q536CKwutP9/RS7VexF1QLZeJocSn9qF3d1TzIbk5uhGXEmc+eTPNnfvtt1kP6Pbss5mXxaXE4eHkYd52VkG6aWTtjP0tx4xRUzRduKAGncrNCamtWHh8IUdvH2XjxY2Pdb/79qm6evXVtH6g5coaVYuVi6/KyFhXWf1Za7VGneytWqVa4ydMyOX7f+88nJkCQV9C92joGga+zQDYeHEj9hp7Nl/arNa9sgC8qkGD36BCP2i+Wh3nQPCuYMZuG6vWs0+N8vRJ6uRcewd02XzHZ+gak5h6EmoO0u3TgnSAv/9Wt//3f1l383gmNV4YPFhdALlyRY3nYArSXR1c1fgOVrp1mKYbM7XK5RWjEdavh/feUy20U6eiprIqUk8FDFEH1ABb2ofcccINdRHRtyW0Pwz1ZoDGPutCZQgITN/vpot+FuuGbYezP6qLhxnm57ZweQH8XQtW+KiMi7tnIXIvbKgHS51hRRE4P/2hXmZeOBt1lmJuxahevLq5X/q6deoCzZQpql/+iy+q1vRu3aBDBzVewT//qMettaSbWrynTlWfzfHjc9dHvUqVtP8XLEjLkEq/HFLnYnfPEKTbu5rr7+uv1YW/Dz8Ep8RTqmtJsaYqq8uQlHbhz/Q51SeBIYXy5VW6vGnguqQkSHJKTZ+6/Ls6Dgy6tCA9/goUb6WyQVJdvKgCz4kT1TFvdCxieVxrHNS0cVcXQZ1p8NwOeOEsVOivUtN3vADLPGCZOxwYYv2NMmjVYLWXF6hsAG0chZLUDCrR0RnWTbqdlp22/03Y1R1ijsKmpmqcGHtXiNoLt7NImchKUjicnqIuuJ/+VmXZ5dCWLbB2LXz5JVy+DKGhKmjPypITS/Ce5M23/36buzIKIR7I0xmkX1+pRrqu85O6H/CO5eigwrbEp45eE30IdAk4OGQY1R1Uqrvpxyng/bTBbGzQoVuHKOpalBeqvMDVO1eJTFBnIn9f+Jv2/u3ZdXWXmjc8ao86sa3yrkpTTwpTJzFxF1WLl1ELscfMQUpsUiy/HfmNo7ePqh05+6gUP+09aDBHpRg6pAYvSeHgXQfK9c66oBoNeFRQ/6f2nYvXqTfeFGi4O7oTnxJPzZpqtV9+UbezZ2e92ayCdFPg7+HkkWWQ7u+vbs+nS3xJTlaj6HbrlpZKX7Vq1vu3RRHxERwJPUL/Gv0ffZBu0KqTQ4C7Z4mNUQGSKcUVUHVv757alUIDRp26WEPmViWtVrXEvPmmGtHY3h7++gvVInT4Q9jSSmUrnZpsvTxRB9R2A1KP83vnwE6ll/9z7R+G1B3C5supQXr8FShcU5Xv6EjV+nTvPGFxYYzdPpZJuydxIfqCOvZBBYs+jaHmN9m/J15BFncTDaozraujq/k2UZtoPq5GjVIpwAMHZr3J6qkN92fPqsEOQbWkm7aZXUu6KRg6k8cTVHzyCXTtqtKJ27VL/Rzp4tO+F/a9DltbQeSeh9vR7U3qgo4pfb7k8+r7J4dMFzh27kxbFh6Oyija3VMFV/fOqxRia8J3wN5+UPE16HAUanytyrOzsxosrMsteP6oOpby2bmoc6ToU3h9zes42DlwLuqcuf93+nEI9uxRAdUPP6jp+FatUv9HJ0YzrcM0kj9N5o1abxCdFG3OMjt0CMqWzTC6eA4ULWq9W1LZsuoxk6AgVMaVXbp5CYvUpXZt9e+VK1C/vjqO7XVR4FREfXb3DVAXSW6lXu1KvAFle8AzHwFQIfVnx/R7EhwMd43+gEYF6RsbqKDXqahaFrXXopy//64y7LZuVRcYfv0VdPZFVANN6mcbe9e0GR3KdFfTtl5drH5f976m+r93PA6dr6mGHGtub1Yt+83XQNAX0PA3PKv3xc9PjSdgwc4pLesnYqfK2rm5To1s32oz1PkRmq+Fcq+oixAR/8CF2XBpXtYz1STcgpBq6vPmVVVdUL2b8+lnTV0ohg9Xt87OaVkU1qw5uwZ/b3/WnF2T430IIR5cvgfpCxcupHXr1gQFBdGtWzcO3qcz6f79++nWrRtBQUG0adOGxYsX536nSeEqvdE0VceKIupKZE4ZdOoK/O5XYE8/2DsAEm5gNMLJk7Bhg/q7dg0V3NxYrU5QTwarQbuedAZt3qQTGnTqPf4n9Qfy5nq4/DtVq6oBbSx24V5OtQzHX4VSz5sD13nzoHFj6NxZDTY2btzDFysrx4+rlNbgYHVyER6O1VTOg7cO4mjvyPQD09Gg4dCtQxiNRkLOhzCk7hBKeZZiy+UtqrXBzln1Wzv7o5rn/HzqxMkO7qlX4RvDri4AzDo0i482fsTgdYMxGo0q9c9ogNvpgj57D9ViFrZF9UUvHJTxZVjKcGJtCtLdndJa0uO18TRsqF6ePgddA729oVq1tPt+fqlBeuo23Z2ybkkPClKtvcuXp9X/kSPqRCwn81c/alptusGaDFq4vgqOjYUjI+DctCyft/nSZioXrUyf6n3YfGkz+qz6ZD+sO6dgXSWVXglwYQYlPFXHUdNIzmaupVT/bOeiUH9Olpvcvl39/fwzfPaZCmA//RTY9T91EanhfGi60tw6nokuHtCo4/LSXNjUBE59ze7ru3F3cmd4veEcCT2iLmbZu6kWOFCB2sXZkHSbladXUt+vPm0rtmXZyWUq6HT1g9AQ9fkxBaFZ8a5pcTfRWUUoGdPdTd0oIiNhyBDV8pSV9IO+maRPd3d1dM0ySC9fXnXpWL06bdm2bVZXzbHkZNWy2qePalnt3Rt++gk1ZkV8aufhBnOz30hiGJz5AQ6+pTJ6ri3LYmeR6jvKsRAc/UR1tflvZI7L+swz4OCgjqk9e9QYCYvnXINLv6lW+Wqfqlk8KvS3voFbf6vfhIB3VRB175yagi85HJ79RJ0DXFuifjPy2dnIs9xJvsPC4wtJ1idzNvIshQqp7kLpf+NM2SnpjysnJxWkF3MvhpO9E0XdihKdGE2ZMmp+eUjrmpEbGo31LkMajWrRN6lZE/X5KlovbaFPYwICLIN5AL1jcRVYGw3qIrNrSZWtZe8Kof+ntmOnzslM+x43DmrXVscrds7gmXr1ypQl5+AKXoFw7mfVvzx1kNPRo6FtW/jzT/X/6tXgWLyeGvzSoqU6NUVA46AGyTwwWH03RuxQM584eavvpAwzP5glXAM0akC6c1Phr6pw/DOqV1fH7anUeHnxYtTrNQXQ7VLLn3hLNRgVqgJnvlMZH2d/hF2d4cDQ1OwTI4Tvsr7/a3+qbLnmq6FEWyjdFTz8ra9rhWk6yZz8dibrkvn7wt980/Yb9t/cT+i9bL78RMERfx3CtqkLTnH5/32YGydPwltvqXP7fv1gxAjVxax/f2jWTC3v1k1l1RRU+Rqkh4SEEBwczNChQ1m9ejV16tRh0KBB3MriV+X69esMHjyYOnXqsHr1aoYMGcJXX33Fhg0bcrdjjwqQcBWSU/tyuZXFWgtRlk4Fw4kvofYPEPgZ+L9BeJiGoCCV1qXVqh+z//5DjWR6YZYaIdnvRfVjYDRyKeYS/1v8P4b/NRxDLgfVyW///qsO/PfeU+lsX34JYRcvqde693XVx/n4WNXn+WFEH1I/QvVTm2XtXcHehdat1YA4kyapEWynTIFo19S5iva8ClcWgfYu+476MmCA+sCuWQPLlqm+fI/CN99AvXqqH+qrr6oLAzodmdJYMRo5FHqI23G3+ebfbzCi7p+MOMn1u9dZd3Yd9nb2hJwPUX3s9AmqL1vtKSqVz9FDvQ8Ru8C3uTkbRGfQMW3/NGa+OJPz0efZe2OvakF0LAz7B8F/o+HeBbCzV61a56erk4L7tZiVtJx2K85LNQ+aWr1NqenFilkPSrJimloKoH37nKe7u7mpL98tW9SFl5494bvvIDAQNm9OS7F/nIPfGI0wcqRqMfr8czVV1uTJqPf95JdQ/lWoNAjcy2c55c7GSxu5l3yPKXumEJMUw6HQQ1y9Cl99pbb5ww/qOL96VX3mhg9Xx9y338LevVY3ad35n8HRExrOU/ddSlCvRhQeHmpe7vh4dcFj505U3V9bplp7EjJG8GmsjsStT4ao/VCul+pasauz6nduTaEAILXlKOhzKN0FUBcudAYdg9YNAlBjNRQOgrCtkBytWmlTLT21lGeKPkPlIpVZemqpWuj3orpAemCIus2O3//SWgPtnEgqpi4opA/Sk3RJlCgBTZtmvymT+vUtB35zcVGDLmbsk260ckHTzk6lyJta0jt3Vt93D8NoVBfRHDOOVeXTWH3H3N4KRWpnvQF9CmyoA3HnVath5WFpGQsZeVRSAdHdM1D9SyiWxQWaLHh6qpb+6Gj1Xdq3L2j0qR9q56Iq8N75P/Vnjb2LGkfDoFcXDI59lhZoaWNVavXNtTaR7n4u+hxNyjTh3QbvUrtkbc5Fn6N1a/XaV6xQ68TFpbWqp8+uMBqNxCTGUNilMACFXQoTkxiDRqMyJh5Gr15p/9vbpx3LplHL/f3TDSTn01jdauyh1PPY28P/MlRNimd91f/66p8qNd3RSwXmJdqqejj8gTmAbt5cZfYYDBm6zmT4PcLJG4q3VGNQbG4CN9diNKrvsUwDtxVvrb6L9qbOh6a7p6YkBXUhu9KbUOYlwE59F2hj1Ycm+lDWF5hcSwNGuHdWfYc4F4PEUF54Qf3+N22qLjgMH456nbHHVNBvGvzOuZj6nCTcVIO5GnVw96T6zq05SV3YjD6sMg2sMaSAJvXiRugG2NFBnW/mUMOG6tbUfQey/u3cdmUbidpEbt27RRHXIqw7ty7H+8ktgyFtZozjxyEq6pHtKl9FRqouGcHB6oLkTz9BbOxjLMDeAaoRLCVanZtE/vsYd/5w7t5VY1Pdu6cugs2fD2PHqvP9f/9V54PLlqmsI1NmTkGUr0H63Llz6d69Oz169MDf358xY8ZQokSJLFvH//zzT0qWLMmYMWPw9/enR48edOvWjd9Mw33nVPm+qkVm54vqfrLqPJSoTeTLHV8yfsd4/r2ezcGqsUMF9QbVJ27Hi1w5vJ+TJ1Xw8eKL6gTj+fZGlZ7pWUkNanb2B9jRgRTtPXot70WVolXYcXUH3+/5Pnflz2cjR6oUzilT1FXqzz6D4imr1KitdaeprgORe9VJd05Zm/LDvbzqE33r/9Q68Zfh4q+89ZYKhkeNghIl4KOPQOdcFvw6q33u6QO6exQpao+zM5w+rdLjDQZ1lc2q2BMqmyJ0g0qVvL01tYUvA4NepdgefEudbBwfB2e+49w51T+vQQPVghEUZOUkIdXBWweZ2mEq19+/zpt13uTgrYOEnA+hrFdZHO0dqVqsKiHnQzCWaA+FnoUdHVUftjvHU9Pf34FTX8O/veHKHwCsOr2K63ev8/3e77HT2PHDvh/UCVCV4erE6FSw+iJ29oHKw9UX8pEP1Ymqo6f1ggL4dVGtBAAlOxDvqvofpg80TNNIffONOpkDdcHCJ4vzd1BBB6j36MUXLdPds2tJB3VRyNFRDaa0bJlqcRsxQl1VrVdPzUkbGJj1vvNaZKR67a1aqQB65Ej4+GNUuqSzjzqOry2FnZ2sXrgyGo1svLiRZuWaEVgskCpFq7Dx4ka6dlU/MOPGqQtiH36o0jjHjFGfu48/Vsd+w4bkfMqcwtVVf8uY1FzMCzNxvHeQUaPUyXCJEurv669RJ60YVSC077XUVp3M223ZUp2wf/SRuhgWEgILFjmDbws1PkRKjJqbOGqf9TIVa6YyNv55SY2wnNpKtvnSZlqVb0Wzss2o51dP9Uuv8pZqOfo7yByg3U6IYseVHZyLPsfh24c5FnZMzTddaaj6DFyYqU7+nbL4QAI4F1GtUAAVB5CoUS16GUd3BxUsO6QOf+Hvn8VgWqh1vkmXZT94cOaB44wYSdJZH3lu7NjUPr+oTIVMwXUuubio9PwFC9SFpA0bUjOLyvVWmWXb2qhB3gBr9YxRrwIbl+Iqy+nEBDVwpUn6Y67UCypQ394BTozP3Hc8B1M6jRuX9voBDO5VwLsW/DdKjSHjm81VwXKvqP7H+wemZiM5qXIXra++u2OOWo70nVHUATjykbq4c32luvCbeJsUfQozD85k2v5pXIzOm+aZs5FnGVp3KD88/wOvBr3K2cizvPqq+h155RX1nVauHHh5qQyk4cPVd+D48TD2y0SS9cl4u6jX4u3ibe6jPmaMmmEAVJp6vXpZlcC6Ll3S5vnu2RPzYHTPp8bJFkF48dbqtlRH84Wb8ePTLtw1bQolarRR3Qv29FHT/JlSuCu/pVrXz36vPvvORXFxUWNbmJQsmTpY6DMfqpZnUF1USndVzzctAzQ+dRk4UP02TJsGO3aoz2Gy0VtN9WYwpTppwKeBCvz3DVRZe2Fb1bFZcQCcnqwaCVyz+ICDCrw9q8CubmpslyQ16NuAASrdPiZGBQzOzqgLAK6lVJeSdamD4Pk2V+eGOzup3+KUWJWB4lYari1X56mOhVSKvzWlu6rvw4NvqXKku2hmMBrQG/RWLwKadO6sutb07asaF9q1Sx001Io1Z9ZQ0bsi269up4xXGdacXcO+faqb0x9/qNT5jRstp500MRrV8di7t8qCW79e/VZYy7w7fFjV97x56r6Dw4NlgxQEHTqo9/vjj9Xn+u231ef8sTAaVQaLbyso0V6doxwYmncze6QXe0L9FlyYpf6OfwFh29DpVN2eP68GirQ4P7f2G5Hu3MbOTmUSJSWlNc4kJqpuOnfupB2H0dGQkMUESAWBQ37tOCUlhZMnTzJ48GCL5U2aNOFIFqMOHT16lCYZcrCaNWvGihUr0Gq1OObwLMbgWBhj63/RX/wVAH3FN0kq0YUJ28fj6+5LpSKVmHdkHnZGO6r5VuNu0l0c7R3R6rV4OHvgXvkjNE7F4cgYdeJaphd1KpVn2zY9q1erk2lHR9W3Z0CPbXDxVzj8EWhcoMoI/jq3kYalGtKuQjual2nO/GPzuX33NsXcczZ68rGwY7g6uFK5aGWMRiPbrmyjfOHyXL97nRblWgCq37OPmw9lCpXj0iU12rZWq360tUWPUNilMBW8K6Az6NhxZQcNSjcg+rYHt26lnYD6FNNxybCTuqXqUsi5EJdjLhObFMvChbWYP18FCkWKqC/afq8Oo2SQKxwdo/qJuVdSf1a+hXftUq0BgYHqxNFggGpxOjWSsekbyjRi0nP/or84HzSg1xvAfyg+PnoOHYLvv1cpuh07qhRqfcOFqlvBvbPg14mKZTuyf7+ehQvVIEn29iqgeO1dK9+C59dBYiQUsoO4cxC2A+zcMBapizE1y0KDRp26Fmmk+oIZHUHjBrFn+fFHPXXrqrmmTSeWQ4ak9i01jQKl15OgTaBm8Zq0r9ieku4l6RrQlZWnV5KYnMjnzT6nf83+RCdEM3DdQG7eC6Nki80qTfreWSg3APzfUD/gbhUgdBN4VoNq4zl1ZjdzXpxD/5r9OR91ntFbR3M38S7uVceBo6+6eFGqg2rV1dhBi40qVdi9gpoD21RP7u6Z66zOdHWSGjQel9gbdKnSBTvs0Ov11PCtgQYNer2e2rVhxgz13Llz9eaWu/Sv36RePTWgT7Vq4OBooG2FtpTyKIVer6eUeynaV2xPijYFezv7TM9v3FilEn7/vboQ8PHH6gTSz0+NiHz3rloWn5xE6L1QNBoNKfoUvJy9cDX4cuaMBl9fNVoyqM+qo0sKybpk7DR2GIwGXB1dcUiJUS3Ijt6AERJvEu1SluMxVwgqproJnI48TeUildmzx5dly9RnwtVVtfi/N3SBeo/3D1dz+lZ+l6tJOiKu7aOun8oZ3XdjH072TtQpUYef2v2Ej7sPVYtW5WLMRRYu1LNokWpNL1xYtcy89ppKI/3hB3Uio9FAo0ZQzzQhsZeX+bNzLfoyofdCKetVllv3blHIuRCVyw8Eu8Loz88B58HoS3eHIo0ZMUJP1aqwZIkq+8CBoPeoAm32pA5q5AL+gwGHTMeHu7s6EZ09W6WX2turNDN945UqQD4+QR1zQcHciY/mTOQZSnqW5E6SKmc132rYNV2nUlbvnIDSr5BUrg+lL31OcOtgqhStQr2S9fj7wt/oXctBm3/h3HQVMNaZwfEUF16v8TozO6l5oj7c8CGnwk7hH9AJWu1S5XcrC898kP1x/uxocCwCQRNwvnubTpU74aRxQq/X80yRZ2hQqgF6vZ4GDdRJaUiIqm9Hx6yP85dfVimvXl7QvTusWNmKMh5l0Ov1+Lj40KlyJ5K0STjZOWV6/rPPwvbtepKSYNAgPR98YP2k9lSESqGtWkxdTNt2ZRs1i9fE2zVzEPrjj6plb/t29T1csybo7VxVPZ/5VrU6l2xPuFsNzlzaRvNyKhA+Epr6m9Fmj2oJ/O8zsHODyh9m8Z7aQ8utKnPj7nko0wcqvpH9+59hWe3aKsX/xx9VOYcOA71+g2p1PT9bBWbPfmb9TXGvBM/tVb+9lxdD2b5QqitUeEN9n15fC07FoeJwSElQQaKdg7o1GlQWtNEBDHYq2A/bToK9DxPP7MDXw5eyXmX5dve3DK07lGq+1VKrTZ92a9CpC6F2DqnZVAYiE+5wIuo0Lcu3BOB42HHsNHbUKVGHGr410Ov11CtZjwM3DlCokJ69e9UJ/MmTKjBu2lQFnPPmqe8/T0/o0jueThGdKO5WHL1eT0CRAOqXrI9er6doUfV5njtXXUBzcsr6OM2qTn75RT3/yy/THmrcWF1AeiNddVKsDQR8ooLg1IUlS6rPyerVKhvIYET97pz5TmVjVHobKgxS2WHN/w8uzVEXpQPeBb2eN95Q5wZ79qixFJycQE8pqDcPQv+Gap+B0Q7c/dXzz01TQa//GwQH66lXT2VcbduWNg6KvtZU9E7lQZ/6/VS0CTRcBOdnqOngSvdQr8HDHzwDIWyXOser80sWfbnsocUWdUxGHoQSHaDCazg769m9W71vsbFq1HS9nRu0/hfOTIKUOyrA9mmplp37CcL3gV93dYGg8gcqmDn4rrrIFPBJ9sf5pTlwdjqUeAF9qa5sPLOevTf2UsqzFHeS7vBStZco61mRCxcgIrUR39FRfcZ27VIzohw8qC4svPhiul2lOyYMBgM/tf+Jtv5tOXr7KBN3T8TeIRkvLwfzgKMaDWh1Og7eOM7NezfNFyRrl6hNjRqFMRjUOnZ2qm6tvaRKlVR3qTt31IOnTunNU/oBFr9xVmnvwd3T4OQDaEB3B72DNzsiL1KvVD08nT25dfcWV2Kv0Lhs40xPD4sL43zUeZqWUylT+27so6SHH3G3SqPTqd9fe3v1Gm7qj+Dl4kVF74rEpcSx/+Z+mpUIwjHhqrrwaUhJHTCwNHHGsub33fRezZ2rWoG//TZt9oZevaCw8211YVOfBI4eGD0qk6BxQqvXYqexw4hRNWYkx3H1zlVz1yk/Tz/u3i7G1q1QuXLaOU7RoqoxwcFBnStrNOrCkU/LbaoL0X9jwcEbKg4mPjmRc7dP4unsSXxKPC4OLvgX8ef6nevcSb6DU2o34YreFdl3cx9Vi6jfnRuxN7h+7zqNyjSCexdVVw6DVl1kcvIB7wbq82rvCcmn0MXeYNJMdZ74wgsqFrh3L132WVa/EQB6Pa6ualDK+fPVubajo7qYOWOG+t4JDlbdOTw9LS94p5eUpC4OmLoWFS9ufcykm3dvcv3OdQKLq9afQ6GHCCwWSLIumTvJd3B2cEZv0FPKsxTHw45TxqsMfoX8iE6I5njEcSp6VyT0bihVilbBYDTwX/h/POf/XOYdWaExZneZ7REKCwujefPmLF68mNqmUUaAGTNmsGrVKqsp7O3bt6dr164MGZI20ubhw4d55ZVX2LVrF76+vjna9+7du3F5LHPbCCGEEEIIIYQQUMfUb+g+8q0l3USTIaXBaDRmWna/9a0tz06jRo0wGo3o9XqOHz9OUFAQ9qY83fyWsSXZCp1BR+2ZtRnddDQzDs2gd1BvBtcZjN6gZ8bBGSRoE3iz7psUci7EsmWqz6qp1dreHl7smkDQ9CC+a/8dE3ZN4J3679C3Rt8s97f4+GJORZyi+7PdqVmy5oO9nvRyOa+QTdaTsErq6v6+2/Md/17/l0RdIp2qdGJI3Sym93mEpJ4KhizrKUNrUt+VfalUpBJrz61lXItxdAroZGVruRMRH0Gv5apj8g/P/0BQ8fsMMvmU05csyfG//872M7Xr6i6G/DWEmsVrElA0gLEtxz7mUoqn+btPr1eZZikpqmXRYhwRG5Orerpf6zrwz9V/eHXVqyzqtojuS7tzbOixLLNXDUYDy04uQ2fQ8XLgyzjYZR0qbbm0hT+O/UGlIpUY03zMfV/XfemTIHSjmjLUwVM18fo0BtecNUI+br/s/4VGTo04xSl61+ytMpHunFJdP4wG1dWoWGPV9czG3LypBqE2jdXi4aHGPHpY686uI+RCCPX96jOg5oCH2la+Bene3t7Y29sTaZoIM1VUVBQ+WXRm9fHxIcKUq5MqOjoaBwcHCpvyOnLAzpSbk8re3t52vqxzMJmpvb097zZ6l9HbR5OiT2FA7QHm1/BOo3cs1u3Vy3IAGMWTN+u9ySdbPyFFn0KfGn2yff2v1nz1AV5IqtxMznofNlVPIltSV1l7r9F7nIo6hY/GhyH1hqiU/nwi9VQwZKqn+Pi0ASCAr9t+Tc9lPQkqEUSXql3yZJ8lCpVg++vb82RbT4XUjo/ZfaZaVmzJiKYjuHH3Bp80+UQ+e/noafzus7fPfqwYW5RtPaVvnDP108wiObhFxRZULFqR3qt706dGH0oUKpH1PrGnd41spqdNp13ldrSr3C5H6+aIvTuUe8hRHx+jofWHcvToUXrX7J1aT/ZQtEZ+FytHypZVf3mtS9UuefY7nG9BupOTE9WqVWP37t20bdvWvPzff/+lTZs2Vp9Ts2ZNtmWYh+aff/4hMDAwx/3RnxSv1XzN3C/dNLhRbrxV/y3cndypVaIWjvZP13snRH5ytHdkbuf7THUlRHYynIiWL1ye/YP251NhnnKmQMHUX9LLK9uL06/Xev0xFEqIp0Aue+v+2ulXDt46SFv/tvdfWQgbkK/p7gMGDGDEiBEEBgZSq1YtlixZQmhoKL1Sm36nTJlCWFgYkydPBqBXr14sXLiQ4OBgevbsyZEjR1ixYgVTpkzJz5eRL5zsnbJNUb8fT2dP3mnwzv1XFEIIIYR1pkBBr1e5k7nsziWEeDwqF61M5aKV87sYQuRYvk7B1rFjR0aNGsX06dPp3LkzBw8eZNasWfilzmkTERFBaGioef0yZcowa9Ys9u/fT+fOnZk+fTpjxoyhffv2+fUSnijh4eEMGTIErWk+gwIuLCyMYcOGPTGv50ln68fftGnT2LhxY34XQzyB5NgXTwM5zoUQIufyfeC4Pn360KdPH6uPTZw4MdOy+vXrs2rVqkddrHzzzz//sGjRIhISEoiPjyclJQUAZ2dn3Nzcshwgz2g05mp9E1dXV/r06UOTJk1YuHAhL774YrZdB65fv87UqVO5dOkSPj4+DBgwgAYNGjzQ+mfOnGHx4sVcuHABo9FI5cqVGThwIGXTdRIxPf/ChQsUK1aM119/3WJ/e/fuZeHChYSHh+Pp6ckLL7xA166qP0/x4sUJCAjg77//5n8Wk7qKrOTF8Wc0GnFxcbnv8Zf+2AOsHn+LFy8mJCSElJQU6tevz7Bhw3C1NpcGsGzZMjZu3Mjdu3dxdnamdu3avPHGG3h4qDl0Fy1axNKlS3FycjI/5+2336ZZ6kghCQkJzJw5k8OHD6PT6ahSpQpvvvkmpUqVAqBHjx588skntGrV6qnrXvM0eFTfvRmPc2uyO/YTExNp2LAhw4cPz/LYnzt3Lvv37yc6OhpPT0/at29Pjx49zI8PHz7cYjwXnU6Ho6MjS5YsASA2NpbZs2fz33//AdC6dWv69+9v7gsqx/6T43GfY6QXFxenxtR5913zZyI33/HpyzJy5EhOnz7NokWLzN/xOXn89OnT/Pbbb0RERODs7Mzzzz9vPgeV41wIYUvyPUgXllatWsWNGzcyLU9KSiIpKSnH28nN+qtWrSIoKIi9e/fyxhtvZLmeTqfjyy+/pEWLFkyYMIH//vuPb775hh9++MEcyORm/fj4eNq0acPHH3+Ms7MzS5YsYdy4ccyePRt7e3vz85s1a0b37t0xGo1MmTLF/PyYmBgmT57MO++8Q4sWLbhy5QqjR4+mbNmy5ukNWrduzbRp0yRIz6HHffytWrWKJk2acPfu3UzH3+bNm9m0aRMTJ07Ey8uLb775hlmzZvHuu+9a3VaTJk3o2LEj7u7uJCQkMH36dObOncvbb79tXqdevXqMGWN9BNaFCxdy8+ZNfv75Z1xdXZk9ezbfffcd3377LaAu+pQqVYrdu3fTsmXLHL8XomB4lMe+6Ti3Jrtj/6uvvuL69ets3Lgx22Pf0dGRkSNHUrp0aUJDQxk3bhyenp48//zzAPz8888W63/55Zd4e6fNo/79999TuHBh5syZQ1xcHOPHj2fFihX07NkTkGP/SZIf5xgZJSQksGrVKhITE3P1HW8SEhKS7aBvWT1++fJlJk6cSIcOHejevTuARbamHOdCCFuSr+nuIrNu3bpRunRp7Ozs8PDwoGjRohQtWhQPDw/s7OzM9zP+5XZ901/p0qXp2rUrR44coXTp0nh6emZZtpMnT3Lv3j1efvllnJycqFevHoGBgWzfvv2B1q9Tpw7NmzfHw8MDR0dHunXrRmRkpLnFx/T8nj174uDgkOn50dHRALRs2RKNRkOFChWoVKkS165dM5fh2WefJSoqiuvXrz985TwF8uL48/T0zNHxZzr2AKvH3+bNm+nUqRN+fn54eHjw6quvsnPnTpKTk62WvVSpUribBm9CTct469atHL/2sLAw6tevT6FChXB0dKRly5YWxxJA9erV2b9fBuh6Ej2q7970x7k19zv2XVxceOWVV7I99l999VXKlSuHvb09pUuXplGjRpw+fdrqutHR0Rw6dMg8YGtSUhJHjx7llVdewdnZmaJFi/K///0vU9qvHPtPhsd9jpF+fXt7e4vPRG6/4wEiIyNZvXo1AwZYn9oou8eXLFlC27ZteeaZZ3B0dMTFxYUKFSpYrCPHuRDCVkhLuo1p0qQJNWrUoHfv3nz77bfmFupbt24xZMgQpk2bZhGIgEohu9/6y5cv59SpU4wda31e1rlz51K6dOlsy3blyhXKli2Lg0PaYVOhQgWuXLmSJ+ufOHECd3d3ihUrlqPnV6hQgcDAQLZs2ULLli25cuUKV65csfhxdnBwoGTJkly6dIkyZcpk+/rEwx9/xYsX5+jRo/j6+jJ8+PAcH3+XLl3KdPxduXLFPIgkqPrWarXcunUr04mVyY4dO5g+fTqJiYk4OzszYsQIi8ePHTtGnz598PT0pEmTJuYLSAAvvPACS5cupW3btri5ubF161bq1q1r8fyyZcuyY8eOHLyToqB5lN+9W7ZsybIlPa+OfROj0cjJkyfN3Tgy2rJlC2XKlCEgIAAAg8GA0WjEmG6kZKPRSHh4OPHx8ebXLMf+kyE/zzFiY2N5//33zcumTp2a6+N8xowZ9OrVi0JZTPKd3eMnT56kTJkyzJw5k4SEBCpVqsTgwYMtMgHlOBdC2AoJ0m1QYmIigEU/KtOPZmJiYqYf0Jys/9JLL2W7z/j4+Pv2A7O2b3d3d/P+H2b98PBwpk+fzuuvv25OU7vf8+3s7GjVqhWzZ89m6tSpGI1G+vfvT8WKFS2e4+rqSnx8fLavTaSxleMvKSnJYl8ODg44OztnebwBtGjRghYtWhAREcGmTZsoXry4+bGmTZvSrl07ihQpwvXr1/nuu+9ISkpi0KBBAJQvXx43Nzf69++PnZ0dpUqVYvz48Rbbd3V1JS6b6ZVEwVaQj32TBQsWkJycTIcOHTI9ZjQa2bx5My+88IJ5mZubG4GBgSxatIhhw4YRFxfHunXrMr1mOfafHAX1ON+1axeJiYm0adOGsLCwXD9+7949Nm/eTM+ePWnVqhV//vknEyZMYOrUqebzDjnOhRC2QtLdbZDphyx9YGn631ogndv1rckYPJ88eZKePXua/0zbSkhIsHheQkJClvvI6fqRkZF8+umndOzY0ZyCmZPn//fff8yYMYNRo0axcuVKZsyYwfbt2/m///s/i+dYO+kQWcuL489Ubw96/AG4uLhY1L9eryc5OTlH2yxWrBj16tVjwoQJ5mVly5bFx8cHOzs7ypUrR9++fdm1a5f58UmTJuHm5saiRYtYtmwZzz//PKNHj7ZIvUxMTMw0SJF4ctjCdy88+LG/bNky/vnnH7744gtcXFwyPX7ixAkiIyMz9bf98MMP0Wq1vPnmm3z22Wc0b94cjUZjcazLsf/kKIjHeVxcHPPmzWPYsGFWt3+/x01lbd26Nb6+vjg6OtKnTx9u3brFzZs3zevIcS6EsBUSpNsgDw8PfHx8uHz5snnZ5cuX8fHxsRps5nZ9aypWrGgxmEy1atVYunSp+Q9US+O1a9fQ6XTm9S5dukS5cuWsbjMn60dFRTFmzBhatWplvhiQ0+dfunSJKlWqEBQUhJ2dHSVLlqRx48YW/cl0Oh2hoaGZWtdF1mzh+ANV/5cuXTLfv3TpEo6OjlYHKbRGp9MRHh5ucfykl3FU4kuXLtGhQwfzGAmdOnUiIiLCol/6tWvX7ptuLAouWz32L1++fN9jf/ny5WzYsIEJEybg4+NjdZ2NGzfSsGHDTKnARYsWZeTIkcyfP58ZM2bg6upKpUqVLAJ9OfafHLZ6nGf3HX/58mViYmIYNWoU/fr148MPPwRg6NCh/Pvvv/d93LS/+5HjXAhhKyRIt1Ft2rRh6dKlxMTEEBMTw7Jly2jXrl2erZ9RrVq1uHHjRrZpXtWqVcPDw4OlS5ei1Wo5ePAgJ06coFWrVg+0flRUFKNHj6ZZs2a88sorWT5/+fLl6HQ6Dh06ZPH8gIAAzp8/z6lTp8x9KPfs2WMRkJ85c4YiRYpIf/RcepjjLy4ujhUrVjz08demTRvWr1/PrVu3iI+PZ9GiRTRv3hxnZ2er2/j777+JjY0F4Pbt2/z+++9Ur17dPKbBnj17uHv3LgA3btxgwYIFNG7c2Pz8gIAANmzYQEJCAnq9npCQEJycnChZsqR5nePHj1OvXr0cvy5R8NjCd2/6Yz8pKYnFixdne+yvWLGCkJAQJkyYgK+vr9V14uLi2LNnD88991ymx0z71+v1HD9+nKVLl9K7d2+LdeTYf7LY2nF+v+/4Z555hjlz5vDjjz/y448/8vnnnwNqqt66deve93GA9u3bs3XrViIjI9HpdPz555+UKlUKPz8/837kOBdC2AqNMf1oMcJm6HQ6fv31V/MAJi1atGDQoEHmflPTp08HMKd23W/9pUuXcurUKcaNG5flPr///nsqV67Miy++mOU6165dY9q0aVy6dImiRYsyYMAAGjZsaH68Z8+efP7551SrVu2+6y9evJjFixdnSsvM6fNBtQytXr2aqKgoXF1dadiwIa+//rp5MLCffvqJcuXK0blz5+zebpGBrRx/6efQrVevHsOGDcPNzc3qNidOnMjJkydJSkrC09OTOnXq8Oqrr+Ll5QXAt99+y5EjR0hJSaFw4cI0bdqUXr16mU8IIyMj+fXXXzlx4gR6vR4/Pz/69u1LjRo1ADVuwscff8zs2bMt5loXT5aCeOz/73//w8HBwWKQzapVq1rsMyQkhJUrVzJ79uxMWSQhISEsXryYxMRESpUqRe/evS2+Z+XYf/IUxOM8vbCwMAYNGmR1nvTsHl+xYgXr1q0jOTmZKlWq8Oabb5pb7uU4F0LYEgnShVlYWBiff/45U6dOxdHRMb+L89DCw8P5/PPP+emnn56I1/Oks/Xjb9q0aVSuXJn27dvnd1HEE0aOffE0kONcCCFyToJ0IYQQQgghhBDCRkifdCGEEEIIIYQQwkZIkC6EEEIIIYQQQtgICdKFEEIIIYQQQggbIUG6EEIIIYQQQghhIyRIF0IIIYQQQgghbIQE6UIIIYQQQgghhI2QIF0IIYQQQgghhLAREqQLIYQQQgghhBA24qkN0hcuXEjr1q0JCgqiW7duHDx4ML+L9FSbOXMm3bt3p1atWjRq1Ihhw4Zx6dIli3WMRiNTp06ladOmVK9enb59+3L+/Pl8KrEAVW8BAQF89dVX5mVST7YjLCyMjz76iAYNGlCjRg06d+7MiRMnzI9LXeU/nU7H999/T+vWralevTpt2rRh2rRpGAwG8zpST/njwIEDDBkyhKZNmxIQEMDmzZstHs9JvaSkpPDll1/SoEEDatasyZAhQ7h9+/bjfBlPvOzqSavV8s0339CpUydq1qxJ06ZNGTFiBGFhYRbbkHp69O73eUpv7NixBAQEMG/ePIvlUk+PR07q6uLFiwwZMoQ6depQq1Ytevbsya1bt8yPS109vKcySA8JCSE4OJihQ4eyevVq6tSpw6BBgywOLvF47d+/nz59+rB06VLmzp2LXq9n4MCBJCQkmNeZPXs2c+fOZezYsSxfvhwfHx8GDBhAXFxcPpb86XXs2DGWLFlCQECAxXKpJ9tw584dXnnlFRwdHZk9ezZ//fUXI0eOpFChQuZ1pK7y3+zZs/nzzz8ZO3YsISEhfPzxx8yZM4cFCxZYrCP19PglJCQQEBDA2LFjrT6ek3r56quv2LRpE99//z2LFi0iISGBN998E71e/7hexhMvu3pKSkri1KlTDB06lJUrVzJt2jSuXLnC0KFDLdaTenr07vd5Mtm8eTP//fcfvr6+mR6Teno87ldX165do3fv3lSsWJEFCxawdu1ahg0bhrOzs3kdqas8YHwKvfTSS8axY8daLHv++eeN3377bT6VSGQUFRVlrFKlinH//v1Go9FoNBgMxiZNmhhnzpxpXic5OdlYp04d4+LFi/OrmE+tuLg4Y7t27Yy7d+82vvrqq8YJEyYYjUapJ1vyzTffGF955ZUsH5e6sg2DBw82jho1ymLZW2+9Zfzoo4+MRqPUk62oUqWKcdOmTeb7OamXu3fvGqtVq2b866+/zOvcvn3b+Mwzzxh37tz5+Ar/FMlYT9b8999/xipVqhhv3rxpNBqlnvJDVvV0+/ZtY7NmzYznzp0ztmrVyjh37lzzY1JP+cNaXb333nvm3yhrpK7yxlPXkp6SksLJkydp2rSpxfImTZpw5MiRfCqVyOjevXsAeHl5AXDjxg0iIiIs6s3JyYl69epJveWD8ePH06JFCxo3bmyxXOrJdmzdupXAwEDeeecdGjVqRJcuXVi6dKn5cakr21CnTh327t3L5cuXAThz5gyHDh2iRYsWgNSTrcpJvZw4cQKtVkuTJk3M6xQvXpzKlStL3eWjuLg4NBqNOatI6sk2GAwGPv74YwYOHEjlypUzPS71ZBsMBgPbt2+nfPnyDBw4kEaNGtGjRw+LlHipq7zx1AXpMTEx6PV6ihYtarHcx8eHiIiIfCqVSM9oNBIcHEydOnWoUqUKgLlurNVbZGTkYy/j0+yvv/7i1KlTfPjhh5kek3qyHdevX2fx4sWUL1+eOXPm0KtXLyZMmMDq1asBqStbMWjQIF544QU6dOhAtWrV6NKlC/379+fFF18EpJ5sVU7qJTIyEkdHR/PFZmvriMcrOTmZb7/9lhdffBEPDw9A6slWzJ49GwcHB/r162f1cakn2xAVFUVCQgKzZ8+mWbNm/Pbbb7Rt25a33nqL/fv3A1JXecUhvwuQXzQajcV9o9GYaZnIH+PHj+fcuXMsWrQo02PW6k08PqGhoXz11Vf89ttvFn2PMpJ6yn9Go5HAwEA++OADAKpWrcqFCxdYvHgxXbp0Ma8ndZW/QkJCWLt2LVOmTKFSpUqcPn2a4OBgfH196dq1q3k9qSfb9CD1InWXP7RaLe+//z5Go5Fx48bdd32pp8fnxIkTzJ8/n5UrV+b6XFzq6fEyDWrapk0bXnvtNQCeffZZDh8+zJ9//kn9+vWzfK7UVe48dS3p3t7e2NvbZ7qSExUVhY+PTz6VSph8+eWXbN26ld9//50SJUqYlxcrVgxA6i2fnTx5kqioKLp160bVqlWpWrUq+/fvZ8GCBVStWtVcF1JP+a9YsWL4+/tbLKtYsaJ5gEz5TNmGyZMnM3jwYF544QUCAgLMLekzZ84EpJ5sVU7qxcfHB61Wy507d7JcRzweWq2W9957jxs3bvDbb7+ZW9FB6skWHDx4kKioKFq1amU+t7h58yaTJk2idevWgNSTrfD29sbBwSHT+YW/v7/5/ELqKm88dUG6k5MT1apVY/fu3RbL//33X2rVqpVPpRJGo5Hx48ezceNGfv/9d8qUKWPxeOnSpSlWrJhFvaWkpHDgwAGpt8eoYcOGrFu3jtWrV5v/AgMD6dSpE6tXr6ZMmTJSTzaidu3a5n7OJleuXMHPzw+Qz5StSEpKytRyZG9vb25xkHqyTTmpl8DAQBwdHS3WCQ8P5/z581J3j5EpQL969Srz5s3D29vb4nGpp/zXuXNn1q5da3Fu4evry8CBA/n1118BqSdb4eTkRFBQULbnF1JXeeOpTHcfMGAAI0aMIDAwkFq1arFkyRJCQ0Pp1atXfhftqfXFF1+wfv16pk+fjru7u7m/n6enJy4uLmg0Gvr168fMmTMpX7485cqVY+bMmbi4uJj7bopHz8PDwzxOgImbmxuFCxc2L5d6sg39+/fnlVdeYcaMGXTo0IFjx46xdOlSxo8fDyCfKRvRqlUrZsyYQalSpczp7nPnzqV79+6A1FN+io+P59q1a+b7N27c4PTp03h5eVGqVKn71ounpyfdu3dn0qRJeHt74+XlxaRJk6hSpUqmQTfFg8uunnx9fXnnnXc4deoUM2fORK/Xm88vvLy8cHJyknp6TO73ecp48cTR0REfHx8qVqwIyOfpcbpfXQ0cOJD333+fevXq0aBBA3bt2sW2bduYP38+IHWVVzTGp7SDwMKFC5kzZw7h4eFUqVKFUaNGUa9evfwu1lMr41zbJsHBwXTr1g1Qre3Tpk1jyZIl3Llzhxo1ajB27NhMQaN4vPr27cszzzzDmDFjAKknW7Jt2za+++47rly5QunSpRkwYAA9e/Y0Py51lf/i4uL48ccf2bx5M1FRUfj6+vLCCy8wfPhwnJycAKmn/LJv3z6rg1h17dqViRMn5qhekpOTmTx5MuvXrycpKYlGjRrx+eefU7Jkycf5Up5o2dXTW2+9RZs2baw+b/78+TRo0ACQenoc7vd5yqh169b069fP3O8ZpJ4el5zU1fLly5k1axa3b9+mQoUKvP322zz33HPmdaWuHt5TG6QLIYQQQgghhBC25qnrky6EEEIIIYQQQtgqCdKFEEIIIYQQQggbIUG6EEIIIYQQQghhIyRIF0IIIYQQQgghbIQE6UIIIYQQQgghhI2QIF0IIYQQQgghhLAREqQLIYQQQgghhBA2QoJ0IYQQQgghhBDCRkiQLoQQQjzB+vbty1dffZXfxchTI0eOZNiwYeb7T+JrFEII8fRyyO8CCCGEELYqIiKCmTNnsmPHDm7fvo2npyflypXjf//7H126dMHV1TW/i3hfU6dOxcHh4X7ur169yowZM9i9ezfR0dH4+vpSs2ZNBgwYQFBQUB6V9MHlxWu0ts3NmzezZs2aPN2uEEIIcT8SpAshhBBWXL9+nVdeeQVPT0/ef/99AgIC0Ol0XLlyhRUrVuDr60ubNm3yu5j3Vbhw4Yd6/vHjx3nttdeoXLky48ePp2LFisTHx7NlyxYmTZrEH3/8kTcFtUKr1eLo6Hjf9R72NQohhBC2RNLdhRBCCCvGjRuHvb09K1asoGPHjvj7+xMQEED79u2ZNWsWrVu3Nq977949PvvsMxo1akTt2rXp168fZ86cMT8+depUOnfuzOrVq2ndujV16tTh/fffJy4uzrxOSkoKEyZMoFGjRgQFBfHKK69w7Ngx8+P79u0jICCAXbt20aVLF6pXr06/fv2Iiopix44ddOjQgdq1a/PBBx+QmJhofl7GVPCUlBQmT55MixYtCAwMpF27dixbtszqe2A0Ghk1ahTlypVj0aJFtGzZkrJly/Lss8/y1ltvMX36dPO6Z8+epV+/flSvXp0GDRrw2WefER8fb37cYDAwbdo0mjdvTmBgIJ07d2bnzp3mx2/cuEFAQAAhISH07duXoKAg1q5di16vJzg4mLp169KgQQMmT56M0Wi0KGfG19i6dWtmzJjBqFGjqFWrFi1btmTJkiUWz/nmm29o3749NWrUoE2bNvzwww9otVoAVq5cybRp0zhz5gwBAQEEBASwcuXKHNW1EEII8bAkSBdCCCEyiImJYffu3fTp0wc3Nzer62g0GkAFsoMHDyYiIoJZs2axcuVKqlWrRv/+/YmNjTWvf+3aNbZs2cKMGTOYOXMmBw4cYPbs2ebHJ0+ezIYNG5g4cSKrVq2iXLlyvPHGGxbbAJg2bRqfffYZf/75J7dv3+a9995j/vz5TJkyhVmzZrF7924WLFiQ5WsbMWIEISEhfPrpp/z999988cUXuLu7W1339OnTnD9/ntdffx07u8ynDIUKFQIgMTGRN954Ay8vL5YvX84PP/zAv//+y5dffmled/78+cydO5dPPvmEtWvX0rRpU4YNG8aVK1cstvntt9/St29fQkJCaNq0Kb/99hsrVqzgq6++YtGiRdy5c4dNmzZl+fpM5s6dS2BgIKtXr6Z3796MGzeOixcvmh93d3cnODiYv/76izFjxrBs2TLmzZsHQMeOHXn99depXLky//zzD//88w8dO3bMcV0LIYQQD0OCdCGEECKDa9euYTQaqVChgsXyBg0aUKtWLWrVqsU333wDwN69ezl37hw//fQTQUFBlC9fnk8++YRChQqxYcMG83ONRiPBwcFUqVKFunXr8r///Y89e/YAkJCQwJ9//smIESNo0aIFlSpV4ssvv8TZ2Znly5dblOG9996jTp06VK1alZdeeon9+/czbtw4qlatSt26dWnfvj379u2z+rouX77M33//zddff03btm0pU6YMjRo1omPHjlbXNwXQFStWzPb9WrduHcnJyUyaNIkqVarQqFEjxo4dy5o1a4iMjARgzpw5DBo0iBdeeIGKFSvy8ccf88wzz/D7779bbKt///60a9eOMmXKULx4cX7//XcGDx5M+/bt8ff354svvsDT0zPb8gA0b96cPn36UK5cOQYNGoS3tzf79+83Pz5s2DBq165N6dKlad26NQMGDODvv/8GwMXFBTc3N+zt7SlWrBjFihXDxcUlx3UthBBCPAzpky6EEEJkwdRabrJ8+XIMBgMfffQRKSkpAJw8eZKEhAQaNGhgsW5SUhLXrl0z3/fz88PDw8N839fXl6ioKEBdFNBqtdSuXdv8uKOjI9WrV7do/QUICAgw/1+0aFFcXV0pU6aMeZmPjw/Hjx+3+npOnz6Nvb099erVy9HrN8n4PmR08eJFAgICLLIOateujcFg4PLly7i4uBAeHm7x+kzrZEwVDwwMNP9/7949IiIiqFWrlnmZg4MDgYGBmVLeM0r/Pmk0Gnx8fMzvN8D//d//8fvvv3Pt2jUSEhLQ6XQW9WNNTutaCCGEeBgSpAshhBAZlC1bFo1Gw6VLlyyWm4JhFxcX8zKDwUCxYsWsppinb/G1Nvp4xkAzYzBsNBozLUu/HY1Gk2m7Go0Gg8Fg9XWlL3dOlC9fHlBB+LPPPpvletbKmb481v7P6nlZdS/ILWvvi+n9Pnr0KB988AFvv/02TZs2xdPTk7/++ou5c+dmu82c1rUQQgjxMCTdXQghhMjA29ubJk2a8Mcff5CQkJDtutWqVSMyMhJ7e3vKlStn8VekSJEc7a9s2bI4Ojpy6NAh8zKtVsuJEyfw9/d/qNeSXpUqVTAYDBw4cCBH6z/77LNUqlSJ3377zWrgf/fuXQAqVarEmTNnLN6rw4cPY2dnR/ny5fHw8MDX19fi9QEcOXIk29fn6elJsWLFOHr0qHmZTqfj5MmTOSp/Vg4fPkypUqUYOnSoOW391q1bFus4Ojpmes15UddCCCHE/UiQLoQQQljx+eefo9fr6d69OyEhIVy8eJFLly6xZs0aLl26hL29PQCNGzemZs2aDB8+nF27dnHjxg0OHz7M999/n2XaeUZubm688sorTJ48mZ07d3LhwgU+++wzkpKSeOmll/LsNZUuXZquXbsyevRoNm/ezPXr19m3bx8hISFW19doNAQHB3PlyhX69OnDjh07uH79OmfOnOGXX35h2LBhAHTq1AknJydGjhzJuXPn2Lt3L19++SWdO3fGx8cHgIEDBzJ79mxCQkK4dOkS3377LWfOnKFfv37Zlrlfv37Mnj2bTZs2cfHiRb744gvzxYEHVbZsWUJDQ/nrr7+4du0a8+fPZ/PmzRbr+Pn5cePGDU6fPk10dDQpKSl5UtdCCCHE/Ui6uxBCCGFF2bJlWbVqFTNnzmTKlCmEhYXh6OhIpUqVeP311+nduzegAtlZs2bxww8/MHr0aGJiYvDx8aFu3brmADUnPvroI4xGIyNGjCA+Pp7AwEB+/fVXvLy88vR1jRs3ju+++45x48YRGxtLqVKlePPNN7Ncv3r16qxYsYIZM2bw6aefEhMTg6+vL7Vq1WL06NEAuLq6MmfOHL766iteeuklXF1dadeuHSNHjjRvp1+/fsTFxTFx4kSio6Px9/dn+vTp5pT6rLz++utEREQwcuRI7Ozs6N69O23btuXevXsP/B4899xz9O/fn/Hjx5OSkkLLli0ZOnQo06ZNM6/Tvn17Nm3aRL9+/bh79y7BwcF069YtT+paCCGEyI7GeL+RV4QQQgghhBBCCPFYSLq7EEIIIYQQQghhIyRIF0IIIYQQQgghbIQE6UIIIYQQQgghhI2QIF0IIYQQQgghhLAREqQLIYQQQgghhBA2QoJ0IYQQQgghhBDCRkiQLoQQQgghhBBC2AgJ0oUQQgghhBBCCBshQboQQgghhBBCCGEjJEgXQgghhBBCCCFshATpQgghhBBCCCGEjfh/cp+Do1uvK5QAAAAASUVORK5CYII=", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.figure(figsize=(12, 2))\n", "ax = plt.subplot(111)\n", "plot_logo(X_attr[0], ax=ax, start=925, end=1100, annotations=seqlets, score_key='attribution')\n", "plt.ylim(-0.025, 0.25)\n", "\n", "plt.xlabel(\"Genomic Coordinate\")\n", "plt.ylabel(\"Attribution\")\n", "plt.title(\"DeepLIFT/SHAP of Marginalization\")\n", "plt.show()" ] }, { "cell_type": "markdown", "id": "d0bb491a-d75f-4ff7-ab2a-23f396597fa6", "metadata": {}, "source": [ "This example encapsulates the strengths and weaknesses of the recursive seqlet function. Starting with the strengths: it looks like it captured all three motifs, and did so quickly and with only one example to build background distributions with. No other false-positive seqlets were called. Further, it called only a total of five seqlets in the entire sequence, with the other seqlets being two low-attribution negative seqlets. As for weaknesses, it looks like it called two negative seqlets with very low attribution value and the spans for all the seqlets are relatively conservative. The negative seqlet calls are likely an artifact of having to construct background distributions for negative seqlets using only a single example where most spans have positive attributions, so a huge dirth of data. When calling seqlets using more examples, this issue should resolve, but one could get around this issue by making sure that the magnitude of the seqlet attributions is above some threshold.\n", "\n", "As for having conservative boundaries... this is a consequence of the recursive seqlet property and the noise in the data. If too many characters have low attribution in a row -- or even negative attributions -- the recursive property may not hold for the shorter spans. Further, calling significant seqlets is going to require good background distributions of attribution values. A single example may not have enough examples of background distributions of all attribution strengths and also for all span lengths. I would expect that this has a saturation property, where having more examples provides a significant boost in seqlet boundary expansion until a certain point where it levels off. \n", "\n", "The reasoning for focusing on finding the cores of seqlets and maybe missing some of the flanks is that seqlet calling is very imprecise and subjective. Where exactly does the YY1 seqlet on the right begin and end? Does it include the `GCGGG` toward the end? What about the subsequent `CTA`? In principle, this can be resolved by clustering together YY1 seqlets to find consensus patterns (as TF-MoDISCo does) but when just calling seqlets with no other information, the exact borders can be difficult to define.\n", "\n", "So, if one views the recursive seqlet algorithm as good at finding cores of seqlets and maybe needs additional flanks on either side -- or maybe you are in a setting where you need the surrounding context of the seqlet for your downstream application -- we can expand the flanks of the seqlet with the `additional_flanks` parameter." ] }, { "cell_type": "code", "execution_count": 5, "id": "dbb128a5-2ccf-4d5a-88de-4605b6adc769", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "seqlets = recursive_seqlets(X_attr.sum(dim=1), additional_flanks=5)\n", "\n", "plt.figure(figsize=(12, 2))\n", "ax = plt.subplot(111)\n", "plot_logo(X_attr[0], ax=ax, start=925, end=1100, annotations=seqlets, score_key='attribution')\n", "plt.ylim(-0.025, 0.25)\n", "\n", "plt.xlabel(\"Genomic Coordinate\")\n", "plt.ylabel(\"Attribution\")\n", "plt.title(\"DeepLIFT/SHAP of Marginalization\")\n", "plt.show()" ] }, { "cell_type": "markdown", "id": "63ecd1c9-ee3a-471a-89c9-a2444ba3778f", "metadata": {}, "source": [ "Here, all we are doing is setting the start and end of the seqlet to have an offset of 5 more than the original call. It looks like the calls may be much better for most purposes, especially for ones where it is okay to have some uninformative positions on either end. Adding additional flanks does not affect the seqlet calling procedure and it does not influence the returned attribution score or the p-values, which remain for the original span. Further, seqlet calls cannot overlap each other, but the flanks of seqlets are not considered an \"overlap.\" Literally, when you add additional flanks, the start and the end just get changed at the very end of the procedure. \n", "\n", "As mentioned before, the recursive seqlet procedure has very few parameters: `min_seqlet_len`, `max_seqlet_len`, and `threshold` (and `additional_flanks`, of course). The minimum and maximum seqlet lengths are used for defining the range of spans to calculate p-values for, with the minimum length (default 4) also defining the smallest span that has to have a p-value lower than the threshold, and the maximum seqlet length just being the maximum size a seqlet can be. Note that this value should probably not be smaller than 4 because of noise in the data. For instance, if you set it to 1, than any individual position that has an attribution that is not very high will ruin the entire seqlet, even if it is just a single weak character in an otherwise-strong motif. Increasing maximum seqlet length will increase runtime and memory usage a little bit but should be set to as large as you think a seqlet might potentially be (default 25).\n", "\n", "Also, note that the threshold will depend a bit on the number of examples that seqlets are being calculated for. A threshold of 0.001 means, definitionally, that there should be at least 1000 examples of spans of that distance (and that the observed seqlet has a higher attribution than all of them). Let's take a look at what happens when we set the threshold to a much smaller value, though." ] }, { "cell_type": "code", "execution_count": 6, "id": "dbdc6cce-88e2-466c-81f4-d056152353e5", "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAA+kAAADtCAYAAADDRDT0AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAACKKklEQVR4nO3dd3QU1dvA8e9m0wshEEKvAaL03jsKiiIgRRRBEUGaXZGiiIgiKCrCDykiCALSe3jpICK9SO8dQghJgPRsmfePm91kk01IICEbeD7n5Gx29u7Mnb07s/PMbTpN0zSEEEIIIYQQQgiR65xyOwNCCCGEEEIIIYRQJEgXQgghhBBCCCEchATpQgghhBBCCCGEg5AgXQghhBBCCCGEcBASpAshhBBCCCGEEA5CgnQhhBBCCCGEEMJBSJAuhBBCCCGEEEI4CAnShRBCCCGEEEIIByFBuhBCCCGEEEII4SAkSBdCiCfcsmXLCAoKsv5VrVqVxo0b07NnT6ZNm0Z4eHhuZ9GGJb9Hjx5NN821a9cICgpi5syZ1mV79uyx2c+Uf++99x5Dhw5N9/WUf0OHDrXZ1ubNm6lUqRIREREAhISEMGrUKNq2bUu1atWoV68e7du35/PPPyckJMT6vkmTJhEUFGR9X2ovvvgiPXv2tPtaREQEVapUyfBzSL0/VapUoW3btvzyyy8kJCSk+9ll1YkTJ3j99depXbs2QUFBzJ49O9206X2GFpMnT7amuXbtWrblMSss35M9e/bk6HZ69uyZpnyDgoKYNGlSjmwvNDSUSZMmcfLkyTSvWb6LQgghHINzbmdACCGEYxg7dizlypXDaDQSHh7OgQMHmDFjBr///js//fQTjRo1yu0sZouPPvqI+vXr2yzLnz8/Tk5OdO/e3brs+PHjjB49Ok36AgUK2Lx3w4YN1KlThwIFCnDz5k06depEvnz56N27N2XLliU6Oppz586xbt06rl69StGiRR96H1atWoXBYABgyZIlVK1a1W46d3d3/vjjDwDu3r3L2rVr+d///seFCxf4+eefHzofAMOHDycuLo4ff/wRX19fihcvnmF6Ly8v/u///o/PP/8cb29v63JN01i+fDne3t5ER0dnS94eROXKlVm4cCHly5d/5NteuHAhRYoUyZF137p1i8mTJ1O8eHGefvppm9e6du1K06ZNc2S7Qgghsk6CdCGEEABUqFDBJthr27Ytb775Jq+99hqDBw9mw4YN+Pv752IOs0fp0qWpUaOG3ddKlSpl/d9S25xReoPBwJYtW/jggw8AWLRoEZGRkSxevJiSJUta0z3zzDP0798fs9mcLfuwdOlSChYsSLFixVi7di3Dhg3D3d09TTonJyebvDdv3pzr16+zbt06hg0bRuHChR86L2fPnqVr1640b948U+lbt27Nhg0bCA4Oplu3btblu3fv5tq1a3Tr1o1FixY9dL4s4uLi8PDwyHR6b2/vdMs7p+XWdosUKZJjNweEEEJknTR3F0IIka5ixYrx2WefERMTw19//WXz2tGjR+nfvz/16tWjatWqdOzYkeDg4DTrCAsLY+TIkTRr1owqVarQqlUrJk+ejNFotKaxNE+fMWMGv/76Ky1atKBq1aq8/PLL7Nq1K8f380Ht2rWLqKgonnnmGQDu3LmDk5MTBQsWtJveyenhf3b/++8/zpw5Q4cOHejWrRtRUVGsX78+0++vXr06ANevX88w3ZkzZxgwYAB169alatWqdOjQgeXLl1tft3Q7MBqNLFiwwNpM/X58fHx45plnWLp0qc3yJUuWUKtWLcqUKZPmPTt37mTAgAE0a9aMqlWr8uyzzzJy5Mg0XQUszbaPHz/Oe++9R926dXn22WcBSExM5LvvvqNx48ZUr16dHj16cOzYMVq1amXT/N5ec/ehQ4dSs2ZNLl++TN++falZsybNmzfnu+++IzEx0SYPkydPpmvXrtSrV49atWrRqVMnFi9ejKZp9/1sUjd3b9WqVbrdLiz5u3z5MsOGDaNNmzZUr16dpk2b0r9/f06fPm2zT126dAFg2LBh1nVYtmWvubvZbGbGjBk899xzVKlShYYNGzJkyBBu3rxpk65nz568+OKLHDlyhNdee43q1avTunVrpk+fnm03pYQQ4kkjNelCCCEy1Lx5c/R6Pfv377cu2717N2+//TbVq1dn1KhR+Pj4EBwczIcffkh8fDwvv/wyoAL0rl274uTkxKBBgyhVqhSHDh3i119/5fr164wdO9ZmW/PmzaNYsWIMHz4cs9nMb7/9Rt++fZk7dy41a9bMlv0xm802NwgAnJ0f7Odww4YN1KhRw1ojXaNGDebNm8e7777Lm2++Sc2aNW2adGc2PxlZsmQJAJ07d6ZIkSJ8++23LFmyhA4dOmTq/VeuXAHSNttP6cKFC3Tv3p2CBQsyYsQI/Pz8WLVqFUOHDuX27dv07duXFi1asHDhQl555RXatm3LW2+9lel96NKlC2+++Sbnz58nMDCQe/fusXHjRr788kvu3LljN881a9aka9eu+Pj4cP36dWbNmsVrr73G6tWrcXFxsUn/7rvv0q5dO7p3705sbCyggtPg4GDefvttGjRowLlz5xg8eHCmm9YbDAYGDBhAly5deOutt9i3bx9TpkzB29ubwYMHW9Ndv36dV155hWLFigFw+PBhxowZQ2hoqE26zJg8ebLNTQCz2cyXX37JtWvXrOu/desW+fPn5+OPP6ZAgQLcvXuX5cuX061bN5YvX065cuWoXLkyY8eOZdiwYQwYMIAWLVoAZFh7PmrUKBYuXMjrr79OixYtuH79OhMnTmTv3r0sW7bM5vsTFhbGp59+Su/evRk8eDAbN25kwoQJBAQE0LFjxyztsxBCCAnShRBC3Ienpyd+fn7cunXLuuyrr76iQoUK/PHHH9YAt2nTpkRGRvLjjz/SsWNHnJycmDRpkrUvtCWoaNiwIe7u7owbN44+ffrY9P01mUzMmjULNzc3AJo0aULr1q355ZdfmDVrVrbsz4cffphm2YYNGyhdunSW1mMymdi0aRPvvPOOdVn79u3Zv38/ixcv5p9//kGn01GuXDmaNm1Kz549KVGiRJr1NG7cON1t1KtXz+Z5XFwcwcHB1KhRw/q5Pffcc6xYsYIrV67YNNe3sNwAuHfvHmvWrGHTpk1UrVrVbo21xeTJkzEYDMyZM8fah7558+bcu3eP//3vf3Tv3p0CBQpYAzV/f/8sNdVu0KABJUqUYMmSJXz22WesXr0avV7Pc889l6bFBsCrr75q/V/TNGrWrEm9evVo2bIlf//9N61bt7ZJ37FjR9577z3r83PnzrFmzRr69u3Lxx9/DKjP3d/fn48++ihTeTYYDLz77rs8//zzgPoeHzt2jDVr1tgE3ylvPJnNZurVq4emacyZM4dBgwah0+kytT2ASpUq2Tz/+uuvuXDhAlOmTLF2p6hbty5169a1pjGZTDRv3pwXX3yRhQsXMmzYMLy9valQoQKgunTcr6zOnz/PwoULee211/jiiy9s8tO1a1f++OMPm+Pozp07zJgxg2rVqgHQqFEj9u7dy+rVqyVIF0KIByBBuhBCiPtK2VT38uXLXLhwgc8++wzApha4WbNmbN26lYsXLxIYGMi2bduoX78+AQEBadKNGzeOvXv32gTpbdq0sQbooPoHt2zZkjVr1mAymdDr9Q+9L5988gkNGjSwWfYgg7nt3buXyMhIa3NqAJ1Ox+jRo3nnnXfYvn07x44dY//+/cyePZuFCxcyffr0NIH37Nmz7da22wse161bR3R0NJ07d7Yu69y5M8uXL2fp0qVpbkDExsZSuXJlm/w1a9aMr7/+OsN92717Nw0bNkzzuXTq1Im///6bQ4cO0axZswzXkRGdTkenTp1YsGABH3/8MUuXLuX555/Hy8vLbvrw8HAmTpzI9u3buXXrlk0z6vPnz6cJ0tu0aWPzfO/evQDWANuibdu2mW5FodPpaNWqlc2yoKAgdu/ebbNs165dTJs2jaNHj6appQ8PD3/gcR2mT5/On3/+yZgxY2z6/xuNRn777TdWrVrFlStXrAMKgvpsHoSlKX2nTp1sllerVo3AwEB27dpl810rVKiQNUC3CAoKsjuSvBBCiPuTIF0IIUSGYmNjuXPnDhUrVgTg9u3bAIwbN45x48bZfU9kZCSggpKtW7faBIr20lnYC2D8/f0xGAzExsbi4+PzwPthUbJkyXRHQ8+K9evXU7lyZbu148WLF+e1116zPg8ODubjjz9m/Pjx1ubqFkFBQXabnqe8WWGxZMkS3NzcaNq0Kffu3bO+v3jx4ixfvpz33nvP5kaGu7s7f/75JwCurq4UL178vs3vQdWMFipUKM3ygIAA6+sPq3Pnzvzvf/9j6tSpHD9+nM8//9xuOrPZzFtvvcWtW7cYOHAgFStWxMPDA03T6Natm93p5Cz5TLk/kPb75ezsTP78+TOVXw8PjzRl4urqarP9I0eO0KdPH+rVq8fXX39NkSJFcHFxYdOmTUydOpX4+PhMbSu1lStX8uOPPzJo0CC6du1q89p3333HvHnz6Nu3L3Xr1sXX1xedTsfnn3/+wFPtWT6v1J+jZdmNGzdsltn7DFN/NkIIITJPgnQhhBAZ2rZtGyaTyVoD7OfnB8A777xjU4ucUtmyZa1pg4KCrKOfp5Y6CLDcAEi9zMXFBU9PzwfdhWxnNpvZtGlTuvOYp9auXTumT5/O2bNnH3ibFy9e5MCBAwDWPsWp/fPPPza1rE5OTg90QyJ//vyEhYWlWW7p8mD5DjyMokWL0qhRI/73v/9RtmxZatWqZTfdmTNnOHXqFN99951Nze7ly5czvS1LEHn79m2bEe2NRmO23HCwWLt2Lc7OzkybNs0moN+0adMDr3Pnzp2MGDGCTp062TTht1i1ahUdO3ZM0/IiMjKSfPnyPdA2LZ/XrVu30vRbv3XrVraUvxBCiPRJkC6EECJdN27cYPz48fj4+FjnEC9XrhxlypTh1KlT9+3P26JFC7Zv306pUqXw9fW97/Y2bNjAkCFDrAFOdHQ0W7dupU6dOtnS1D27HDx4kLCwsDTNqm/dumW39jEmJoaQkBC7r2WWpQZ+zJgxafqex8fHM2jQIJYuXZrpqdAy0rBhQzZu3EhoaKhNULty5Uo8PDyybaqw3r174+bmxnPPPZduGksfbldXV5vl9vqup8fSZzs4ONimVcf69euzNGjf/eh0OvR6vc0o/vHx8axateqB1nfy5Eneffdd6tevn24XBZ1Ol2bgvG3bthEaGmozzoLl88tMbb6lO8iqVatsmrEfOXKE8+fP079//yzvixBCiMyTIF0IIQSg5rs2mUwYjUYiIiLYv38/y5YtQ6/XM3nyZJsm2V999RV9+/alT58+dOrUicKFC3P37l3Onz/P8ePH+eWXXwB47733+Pfff+nevTs9e/akbNmyJCYmcu3aNf7++2+++uorm5o6vV5P79696d27t3UKqOjoaN599900+d29e7fdacSyI0i9n/Xr11OxYkVriwGLqVOncvDgQdq1a8dTTz2Fu7s7165d488//+TOnTsMGTLkgbZnNBpZuXIlgYGBaZo7W7Rs2ZItW7YQERGR4cjtmTFo0CC2bt1Kr169GDRoEL6+vqxevZpt27bx6aefZku3A1ADAzZp0iTDNOXKlaNUqVJMmDABTdPw9fVl69at7Ny5M9PbqVChAi+++CKzZs1Cr9fToEEDzp49y6xZs/Dx8cnSYG4Zad68ObNmzeLjjz/mlVde4c6dO8ycOTPNDYbMiI6Opm/fvri7u9OnTx+OHTtm83r58uXx9vamRYsW1lHcLdPPzZw5M00NeKlSpXB3d2f16tUEBgbi6elJQECAzU0Yi3LlyvHKK6/w559/4uTkRLNmzayjuxctWpQ333wzy/sjhBAi8yRIF0IIAagpqgBcXFzIly8fgYGB9O3bl65du6YJ+ho0aMDixYuZOnUq3377Lffu3SN//vwEBgbaDM4VEBDAkiVLmDJlCjNnziQ0NBQvLy+KFy9O06ZN0zTH7dGjBwkJCYwZM4bw8HAqVKjAtGnTqF27dpr8/vDDD3b3Y/PmzQ/7UdzXxo0bbQZvs7BMg7Z27VpmzpxJVFQUvr6+VK5cmenTpz/wDYRt27YRFhZG3759003TrVs3NmzYwMqVK+ndu/cDbceiXLly/PXXX/z444+MHj2a+Ph4AgMDGTt2rHV6vUfFxcWFqVOn8s033zBy5EicnZ1p2LAhs2fPTrfZvz1jx46lUKFCLFmyhNmzZ/P000/z888/8/bbbz9ws/DUGjZsyLfffsuMGTPo378/hQsXplu3bhQoUIARI0ZkaV137tyxdjmwV55z5syhfv36jBgxAmdnZ6ZPn05sbCyVKlVi0qRJTJw40Sa9h4cH3377LZMnT6ZPnz4YDAYGDx5s9wYYqCnYSpYsyZIlS5g/fz7e3t40bdqUjz/+WJq7CyFEDtNpKYfsFUIIIXLBtWvXaN26NUOGDKFPnz65nZ0MHTlyhK5du7Jq1SqCgoJyOzviIRw8eJBXX32VH374gfbt2+d2doQQQghAatKFEEKILKlWrRqnT5/O7WyILNq5cyeHDh2iSpUquLm5cfr0aaZPn06ZMmXSjC0ghBBC5CYJ0oUQQgjx2PP29mbnzp3MmTOHmJgY/Pz8aNasGR999JHd6e6EEEKI3CLN3YUQQgghhBBCCAfhdP8kQgghhBBCCCGEeBQkSBdCCCGEEEIIIRyEBOlCCCGEEEIIIYSDkCBdCCGEEEIIIYRwEE90kG42mzly5Ahmszm3syIyIOWUd0hZ5Q1STnmDlFPeIWWVN0g55Q1STnmDlFPOeqKDdE3TMBgMyAD3jk3KKe+QssobpJzyBimnvEPKKm+QcsobpJzyBimnnPVEB+lCCCGEEEIIIYQjkSBdCCGEEEIIIYRwEBKkCyGEEEIIIYQQDkKCdCGEEEIIIYQQwkFIkC6EEEIIIYQQQjgICdKFEEIIIYQQQggHIUG6EEIIIYQQQgjhICRIF0IIIYQQQgghHIQE6UIIIYQQQgghhIOQIF0IIYQQQgghhHAQEqQLIYQQQgghhBAOQoJ0IYQQQgghhBDCQUiQLoQQQgghhBBCOAgJ0oUQQgghhBBCCAchQboQQgghhBBCCOEgJEgXQgghhBBCCCEchATpQgghhBBCCCGEg5AgXQghhBBCCCGEcBASpAshhBBCCCGEEA5CgnQhhBBCCCGEEMJBSJAuhBBCCCGEEEI4CAnShRBCCCGEEEIIB5HrQfq8efNo1aoVVatW5eWXX2b//v3ppt2wYQO9e/emQYMG1KpVi1deeYUdO3Y8wtwKIYQQQgghhBA5J1eD9ODgYMaOHcuAAQNYsWIFtWvXpm/fvty4ccNu+n379tGoUSOmT5/OsmXLqF+/PgMGDODEiROPOOdCCCGEEEIIIUT2y9UgfdasWXTu3JmuXbsSGBjIiBEjKFKkCAsWLLCbfsSIEfTt25dq1apRpkwZPvroI0qXLs2WLVsecc6FEEIIIYQQQojs55xbG05MTOT48eP069fPZnnjxo05dOhQptZhNpuJiYkhf/78Wdq22WxG0zRMJhOA9VE4JimnvEPKKm+QcsobpJzyDimrvEHKKW+QcsobpJwejF6vz1S6XAvSIyMjMZlMFCxY0Ga5v78/YWFhmVrH77//TlxcHM8//3yWtn3s2DEMBoP1+dGjR7P0fpE7pJzyDimrvEHKKW+Qcso7pKzyBimnvEHKKW+Qcsqa2rVrZypdrgXpFjqdzua5pmlpltmzZs0aJk+ezJQpU9IE+vdTpUoVa0360aNHqVq1aqbvaohHT8op75CyyhuknPIGKae8Q8oqb5ByyhuknPIGKaeclWtBup+fH3q9ntu3b9ssDw8Px9/fP8P3BgcHM2LECCZOnEijRo2yvG0nJ9uu+Hq9Xr5ceYCUU94hZZU3SDnlDVJOeYeUVd4g5ZQ3SDnlDVJOOSPXBo5zdXWlcuXK7Ny502b5v//+S82aNdN935o1axg6dCgTJkygRYsWOZxLIYQQQgghhBDi0cnV5u69e/dmyJAhVKlShZo1a7Jw4UJCQkLo3r07ABMmTCA0NJTx48cDKkD/7LPPGD58ONWrV7f2XXd3d8fHxyfX9kMIIYQQQgghhMgOuRqkt2vXjsjISKZMmcKtW7eoWLEi06dPp3jx4gCEhYUREhJiTb9w4UKMRiOjR49m9OjR1uWdOnXiu+++e+T5F0IIIYQQQgghslOuDxzXo0cPevToYfe11IH33LlzH0WWhBBCCCGEEEKIXJFrfdKFEEIIIYQQQghhS4J0IYQQQgghhBDCQUiQLoQQQgghhBBCOAgJ0oUQQgghhBBCCAchQboQQgghhBBCCOEgJEgXQgghhBBCCCEchATpQgghhBBCCCGEg5AgXQghhBBCCCGEcBDOWX3D7du3GTduHLt27SIiIgJN02xeP3nyZLZlTgghhBBCCCGEeJJkOUgfOnQoISEhDBw4kICAgJzIkxBCCCGEEEII8UTKcpB+4MAB5s+fz9NPP50T+RFCCCGEEEIIIZ5YWe6TXrRo0TRN3IUQQgghhBBCCPHwshykDx8+nAkTJnDt2rWcyI8QQgghhBBCCPHEynJz9w8//JC4uDieffZZ3N3dcXFxsXl979692ZY5IYQQQgghhBDiSZLlIH348OE5kQ8hhBBCCCGEEOKJl+UgvVOnTjmRDyGEEEIIIYQQ4omX5SAdwGQysWnTJs6fP49Op6N8+fK0atUKvV6f3fkTQgghhBBCCCGeGFkO0i9fvky/fv0IDQ2lbNmyaJrGtGnTKFKkCNOnT6dUqVI5kU8hhBBCCCGEEOKxl+XR3ceMGUPJkiXZtm0by5cvZ8WKFWzdupUSJUowZsyYnMijEEIIIYQQQgjxRMhykL5v3z4+/fRT8ufPb13m5+fHJ598wr59+7Izb0IIIYQQQgghxBMly0G6q6srMTExaZbHxMSkmY5NCCGEEEIIIYQQmZflIL1FixaMHDmS//77D03T0DSNw4cPM2rUKFq1apUTeRRCCCGEEEIIIZ4IWR447vPPP+ezzz7jlVdewdlZvd1kMtGqVStGjBiR7RkUQgghhBBCCCGeFFkO0vPly8evv/7KpUuXuHDhApqmUb58eUqXLp0T+RNCCCGEEEIIIZ4YDzRPOkCZMmUoU6ZMNmZFCCGEEEIIIYR4smUqSB87dizvv/8+np6ejB07NsO0w4YNy5aMCSGEEEIIIYQQT5pMBeknTpzAaDRa/xdCCCGEEEIIIUT2y1SQPnfuXLv/CyGEEEIIIYQQIvtkeQq2YcOGER0dnWZ5bGysNHUXQgghhBBCCCEeQpaD9BUrVpCQkJBmeXx8PCtXrsyWTAkhhBBCCCGEEE+iTAfp0dHRREVFoWkaMTExREdHW//u3r3L33//TYECBXIyr0IIIYQQIpdoGnzyCVy9mts5EUKIx1ump2CrU6cOOp0OnU5H27Zt07yu0+l49913szVzQgghhBDCMWzfDhMmgMEAEyfmdm6EEOLxlekgfc6cOWiaxhtvvMGkSZPw9fW1vubi4kKxYsUoXLhwjmRSCCGEEELkrg0b1OPq1RKkCyFETsp0kF6vXj0ANm/eTLFixdDpdDmWKSGEEEII4Vj271ePFy+qJu8lS+ZufoQQ4nGV6SDd4saNG9y4cSPd1+vWrftQGRJCCCGEEI5F0+DAgeTnhw9LkC6EEDkly0F6z5490yxLWat+8uTJh8uREEIIIYRwKJcuQURE8vOjR6F9+1zLjhBCPNayHKTv27fP5rnBYODkyZNMnDiRDz/8MNsyJoQQQgghHMPZs7bP4+NzJx9CCPEkyHKQ7uPjk2ZZ48aNcXV1ZezYsSxbtixbMiaEEEIIIRzD9eu5nQMhhHhyZHqe9PspUKAAFy9ezK7VCSGEEEIIB3HtWm7nQAghnhxZrkk/depUmmW3bt1ixowZBAUFZUumhBBCCCGE45CadCGEeHSyHKR37NgRnU6Hpmk2y2vUqME333yTbRkTQgghhBCOISQkt3MghBBPjiwH6Zs3b7Z57uTkRIECBXBzc8u2TAkhhBBCCMeRcmR3IYQQOSvLQXrx4sVzIh9CCCGEEMJBSZAuhBCPTpaDdIBdu3Yxe/Zszp8/j06no1y5crzxxhs0atQou/MnhBBCCCFymSVIL1oUQkNzNy9CCPG4y/Lo7n/++Sdvv/02Xl5e9OrVi549e+Lt7U2/fv34888/cyKPQgghhBAil2hacpA+aRK8+WauZkcIIR57Wa5JnzZtGsOGDeP111+3WV6rVi1+/fXXNMvvZ968ecycOZOwsDAqVKjA8OHDqVOnjt20t27dYty4cRw7dozLly/Ts2dPRowYkdVdEEIIIYQQmRQXB4mJ6v+6dcHVFfbty908CSHE4yzLNenR0dE0bdo0zfLGjRsTExOTpXUFBwczduxYBgwYwIoVK6hduzZ9+/blxo0bdtMnJibi5+fHgAEDeOqpp7KadSGEEEIIkUWWy7tChaBkSbCpS9E0OPQJRBzKlbwJIcTjKMtBeqtWrdi4cWOa5Zs3b6Zly5ZZWtesWbPo3LkzXbt2JTAwkBEjRlCkSBEWLFhgN32JEiX4/PPP6dixIz4+PlnNuhBCCCGEyKL4ePVYuTLodKpfepEiSS+G/Q2nJsCBwbmWPyGEeNxkqrn7nDlzrP8HBgYydepU9u7dS40aNQD477//OHjwIL179870hhMTEzl+/Dj9+vWzWd64cWMOHcrZu7FmsxlN0zCZTADWR+GYpJzyDimrvEHKKW+Qcso7HveyiosDLy8oXRosu1ihQtL/IdvByQsi/oM758CnbK7mNSOPezk9LqSc8gYppwej1+szlU6naZp2v0StWrXK3Mp0ujTzqKcnNDSUZs2asWDBAmrVqmVdPnXqVJYvX8769eszfH/Pnj156qmnHqhP+pEjRzAYDFl+nxBCCCGEEEII8SBq166dqXSZqknfsmXLQ2UmIzqdzua5pmlplmW3KlWqWGvSjx49StWqVTN9V0M8elJOj4Cvb9pld+9meTVSVnmDlFPeIOWUdzzuZXXwILRsCSNGwJAhapnRCM5ORlhRHMxJ7eHL9oLak3Ivo/fxuJfT40LKKW+QcspZDzRPenbw8/NDr9dz+/Ztm+Xh4eH4+/vn6LadnGy74uv1evly5QFSTjnI3qCPD/FZS1nlDVJOeYOUU97xuJZVXJz6mfDwSP5p0OuB6KtgDE9OaIp5qN+OR+VxLafHjZRT3iDllDMyFaSPHTuW999/H09PT8aOHZth2mHDhmVqw66urlSuXJmdO3fy7LPPWpf/+++/tG7dOlPrEEIIIXJKRFwEt2JuUbFgRZx0WR5nVYjHhmXgOG/vVC/EXn7keRFCiCdBpoL0EydOYDQarf+nJ6vN1Hv37s2QIUOoUqUKNWvWZOHChYSEhNC9e3cAJkyYQGhoKOPHj7e+5+TJkwDExMQQERHByZMncXFxoXz58lnathBCCJGRafunMXzLcCI/iyS/e/7czo4QuSbdID360qPOihBCPBEyFaTPnTvX7v8Pq127dkRGRjJlyhRu3bpFxYoVmT59OsWLFwcgLCyMkJAQm/d07NjR+v/x48dZs2YNxYsXz9F+80KI7LdrFxQrpkYLFsIRxRvjbR6FeFKlG6TH3XjkeRFCiCdBlvqkG41GqlWrxooVK6hYsWK2ZKBHjx706NHD7mvfffddmmWnT5/Olu0KIXKYvZY1SZNJXLkCzzwDFSuqAYlyeKxIIR5InDFOPRricjknQuSudIN0w51HnRUhhHgiZKmTnbOzM8WKFcNsNudUfoQQT4BlyyA2Fg4fhn//ze3cCGGfJTi3BOtCPKnSDdIT76hHZy9A7rYKIUR2yfJIOAMGDGDChAncuXMnB7IjhHgSrF6d/P+aNbmXDyEyIjXpQiiWIN3TM9ULiZHqsdbP8PSQR5klIYR4rGV5Cra5c+dy+fJlmjZtSrFixfBMdcZevnx5tmVOCPH4iY+Hv/9Ofv7ff7mXFyEyYg3SpSZdPOHikg4BF5dUL1iauxd+Bgo1hePfPspsCSHEYyvLQXrr1q2zPIq7EEJYnDkDSZNFANZu6kI4HMuAcVKTLp50CQnqMc1UyIl3wLMEeJdRJ3PPEo84Z0II8XjKcpD+7rvv5kQ+hBC5yRIp63Q5HjUnzaIohMOzBOfZObr79XvXmbR3Ej2r9aRyQOVsW68QOckyFFGaIN1wB7wD1f86HRSs9yizJYQQj60s90lv3bo1kZGRaZbfu3eP1q1bZ0umhBCPr7NnczsHQmROTjR3v3L3CuN2juPYrWPZtk4hcprl3m3amvRI8EhRe+5X85HlSQghHmdZDtKvX79ud3T3xMREQkNDsyVTQojH17VruZ0DITLHOrp7NjZ3jzHE2DwKkRfYrUnXNNXc3a1A8jLPko8yW0II8djKdHP3zZs3W//fsWMHPj4+1udms5ldu3ZRvHjx7M2dEOKxc+OGeixSxM4gREI4kEzVpNsboyWDLiOxhlibRyHyArtBuikeNCO45EteJmMWCSFEtsh0kD5o0CDr/0OHDrVdibMzxYsXT7NcCPF4WnV6FcFng/n1hV+zPJCkJUgfNgyqV4fvvsuBDAqRVQnhcOF3qPgu6N2BnBk4ToJ0kRfZbe6uJY0A6uL7yPMjhBCPu0wH6adOnQKgVatWLFmyhAIFCtznHUKIx9Xfl/9m2oFp/Pzcz7g7u9tPlHIwuhTPw8PV0/btoUwZ+PPPnM2rEJly4H24PA/ibkCtnwAVnLvp3bJ14LiYxBibRyHyAvvN3S1Ber406YUQQjycLPdJf++99/D29k6zPDExkRUrVmRHnoQQDi46MdrmMUvvjYbChaFsWRW/t2qllkfGRbLh/AYi4iKyM6tC3J8xBq4tV/+fmwqGe4Bq5u7n4ZetA8dJTbrIi+wG6WaDenSWIF0IIbJbloP0YcOGERUVlWZ5TEwMw4YNy5ZMCSEcm3XwqweoDYyOhnLlkp/Xrase993YR9s/27Ln2p7syKIQmReyHkxJQbMpHm7tAFRNen73/NLcXTzxLA2jnFJeNZotNek+adILIYR4OFkO0jVNs9sHNTQ01GYwOSHE48vaZDeLI1QbjRAfD0WLJi8LTJpi92Fq54V4KBEHbJ9rBjRNI86YFKRnY026jO4u8iLLZZ/N5D6W5u5Oro88P0II8bjLdJ/0jh07otPp0Ol0vPHGGzg7J7/VZDJx7do1mjZtmiOZFEI4lgcNqGOS4pKUQ1pYLv4kSBe55s6RNIuMZiNmzSw16UKQXINuMqVYaAnSdaknTxdCCPGwMh2kP/PMMwCcPHmSJk2a4OXlZX3NxcWF4sWLU6pUqezPoRDC4Txoc/fopPjb3riTEqSLXBNzOc0iS+25n7sf8absGzhOgnSRV6xcCd9/D7NnpxOkW5q7O2X6UlIIIUQmZfrMOnjwYACKFy9Ou3btcHNzAyAqKopVq1Yxc+ZMTp06xcmTJ3Mmp0IIh/Ggzd0tQbqfn53XJEgXuSX+pnr0qQBRZ4Hkadfyu+fnZvTN9N+bziwG6YkxxODt6i3N3YVDS0yEDz6AS5dg0CCoU0ctl5p0IYR4NLLcJ71Tp064ubmxa9cuPvnkE5o0acKff/5Js2bNWLp0aU7kUQjhYB40oJYgXTgcsxESboOrHzx/BEp2BZJr0rO7T3qsIZaCHgWlJl04tLlzVYAOsGEDWMYLtluTrnOG8P3wT1c4+PGjzKYQQjy2stRG6ebNmyxbtoylS5cSFxfH888/j9FoZNKkSZQvXz6n8iiEyE0JtyF0K5ToZG3W+LDN3e3M4ihBusgdCWGApoJzvTsEvQcJt61zo/u5+2V7n/SCnhKkC8e2cqXt8/ikHh/p1qTHh8DVJeBb5ZHkTwghHneZrknv27cv7dq149y5c3zxxRfs2LGDL774IifzJoRwBPvfhZ3d4OAH1kUxiTH4uPpkuclubFJc4mzn9qAE6SJXxCU1ZS+Q1J63YENw8yfOEIcOHfnc8mXv6O6JMRT0KPhA0xcK8VB0urR/dmga/Ptv2rdCekG69EkXQojslukgfefOnXTt2pX33nuPFi1aoNdLHyQhHnt3jsKVv9T/536F6EtomkZ0YjSFvAplOaC2TN/j4pL2NWuQbpAgXeQQe0GKpT+6Vxn16KSHgvWJM8bh7uyOh4uHtVb9QRw4AKdOJT+XmnTh6EJCIDzcdlmGA8fpstxzUgghxH1k+sw6b948YmJi6Ny5M127duXPP/8kIiIiJ/MmhMhtlxck/6+ZIfoC8cZ4NDQKeRbKcm2g5QIvvZr0gh4FpSZdPFqJSb9j7oWSlzm5EGdQQbq7s/sDN3c/cQIaNIDatZMD9VhDLP4e/hKkC4d15ox6dHKC0aPB0xNck6ZCtwnSLQPGaWaEEEJkr0wH6TVr1mTMmDH8888/vPLKK6xdu5ZmzZphNpvZuXMn0dFyYS3EYydsZ5pFlibuhbwKZbm5+/1q0gO8AiRIF4+WOVE9uhWyWWypSXd3dn/g5u7vvQdGo+rmMXq0WhZjiKGgZ0FiDDFo9xkJXojccOGCeuzYEb74Aj77DJIm9MFgSJHQKelEbmn2noqmJXdxEkIIkTVZbqPk4eFBly5dWLBgAatWraJ3797MmDGDRo0a0b9//5zIoxAit0SdTrPIEkQX8sx6c/f71aRLkC4eOWuQ7m+zON4Yj4eLxwPXpF+8CJs3Jz+/nDQVu2V0d7NmJtGU+KC5FsIqwZhARFwEJrPp/okzwdJI8uWX1WPv3sk16TZBt6UvumZ/u717g68vLFmSLdkSQognykN1JCpXrhxDhgxh+/bt/Pjjj9mVJyGEIzDFQ3wo6D2g6QrwKg0kj+ju7+n/wDXpEqQLh2FOVKO6691sFts0d3+AmvSUAXpKlj7plv+FeFjzj86n4PiCHLt1zGZ5SIiaSi0mo9O0nXEaLEF61arqsWRJKK1O/9g0mnRKP0hfswb++EO1JOnfP9X7hBBC3Fe2jPah1+t55plnmDp1anasTgjhCGKvqsdyvaFEB6g5AVDNdV31ruRzy5emT/qOHerRMqduanZHCE4SnRhNYa/CEqSLR8uUCLq0/S+sA8c5exBniMty0/SDB9Wjk1NyLaSmadbR3YEs3+QSwp7I+EibR1DBcadO0KsXdO6smp6T8jusabbPU64vaTUlSiQv8/VVjzbBtrUmPW1z9wkTkv8PD4ebNzO7N/dxcQ5sbgURB7JphUII4ZhkSE4hhH0xSe1zCz+jHot3AHdV0+3l4oWXi5dNQH3+PLzyivr/nXfsr9IyKYTRThfG6MRoCntLkC5yUMrAxPK/OTG5b20KKWvSNTQMZkOaNBk5llSp+dNPcPiw6tObYEpAQ8PfUzWtl5p0kR0i4iJsHgEmT4Y9e9T/69fD/v1ZWF+E+r76+SUvc3dXj3Zr0lM1s4+KSr5hm62izsO+/nBrK/zdHgzp3A0WQojHgATpQgj7LDXpPuXVo5Mz+FYmJjEGTxdPPF08bWoCP/oouVnl2rVw+3baVVqm8UkvSLc0d5cBtcQjkzJIvzgX9g2E66uJM8bh4az6pAM2/dLnz4fu3eF02iEbrK5fV6Ni9+sHTz+tajQtQXkBjwKABOkie9gL0v/80zZNXBZ6bERGQtGittOo2w3SU9akF2kDlT8H1E0pS2upypUzv937Ov0TmJJ2JC4E7p3KOL0QQuRhdnqGCiEEybUUHsWTl+l0xBhi8HL1wsvVy9rc/d49WLcuuVkvJPc/T8lSk25IVSmZaErEYDYQ4BWAWTNbB+0SIselDNJDt8LFWeDmT7zRyVqTDqr5uy++7NwJb7yhbjTt2wfHjycHMCmFhUGtWsmvde4Md5KOFy9XL9yd3bM8haEQ9qQO0u/cgQMP0Ro8MhJ8fGyXeSSdju0G6WaDGtPB2QuAI0fU4o4dYdkyNYBctghZn00rEkIIxyc16UII+0zxakAtVz+bxdGJ0daadEvT9D170gbe9qTX3N2yngCvAJvnQuQ4syE52EghZXN3y3NQ01FZvr8XLqga89QSElST37Jlk5f5+CTXnHs4e+Dh7CE16SJbWPukx6lHy3gI/v6qmXuLFhm82U4XkIQE2xuukHyzKeV4IwZT0nFjtD1f37ihHgcMULXxX35pWyv/QIxxEH0e0EHxl0Anl69CiMebnOWEEPaZE8DFN83VVUxijLVPuqW5u6XWxtIn/cUX7a/ScuGXOqCXIF3kGidn0NI2+4gzxnE96jorTq2wPr95E3buvP8qw8LUo2VEbAtrkO7igaeLpwTpIltExEWQ3z2/tSb94kW1/JNPoHZtmD49uatRZphMyfOiW1iC9Mjksem4diMpSDfcs0l765Z6rFVLPZYta3vDKrP++0+NEB8XB0SfAzQo9xY0WwnVvsn6CoUQIg+RIF0IYZ8pHpzc0iyOMaTok57UXPfaNfXawIHqcfRo+zUn3t7qMT7edrkE6SLXOLmClrYZSJwhjt3XdtNvTT9AzZv+zz/qtUKFVOBTrJj9VVrGYyhSxHZ5rCEWV70rzk7OEqSLbBMRF0HZ/GWJiFdBekiIWm6pQa9QAerUyfz6jMb0a9JTjtJ+/mL6Qbq3NxQsmLwsKzcJAPbuhcaN4c03oV07MN1JGgCizOvqseL74FYoaysVQog8RPqkCyHsM8WrAAbg0jy4cxSKtVOjuyf1SY8zxmEymwgJ0ePsDFWqqFGtAwPtX5RZgvQ7d2yXRydG46RzwsfVBxcnFwnSxaPj5KqavKeSem70OEMcJ0+q/7/9Ft5+W80fbY+lSbCnp+3yGEMMHs6qc6+Hi4dMwSayRWRcJDWK1LA2d7cE0uXLJ6exN25CekwmcE51dWh5v+UGAMDZcy48UxAw2gbpoaFQpsyDN3HXNHjrreSBSLdtg9iwS/gA+CaNROfsAd5lHmwDQgiRB0hNuhDCPnO8GgwI4NoKODkOwvfaNHcHVTsYEqKaM6a8sMuoJj0iwna5ZVo3nU6Hl6uXBOni0XFysRukxxttm3vEGeOs/c+ff149tm0LxYuThqXPeuomw7GGWBJNiXy8/mNCo0OlJl08NJPZxJ34O6omPam5+82bal7zAgUebJ16fdouSZYg/dgxFUTfuAEht9zVeA6patLDwyF//gfbNsChQ2pAxpScTNHg7A1u/vbfZKfLihBC5GUSpAsh7NPMQNpIO8YQg1kzcy/hnvV5SAiUKnX/VWYYpLuqoD/1/OtC5KiUNek1x0P+6oD9mvTr1yFfvuRm7jqd/RpKS4CT+rVYQyxxxjh+3P0joTESpN/PqlVQr54agCxWPiq77ibcRUOzCdLDwtQc5w9ak+3sDImJtsu81OmZ27fhyhU1swHowDU/JNie0BMTs1Zzn9qWLeqxY0dVK//cc4ApFtwL29+pk9/DQjfY2Ahi7YzkKIQQeZAE6UII+/TudmsYoxOjWXBsAfV+q2d9fu9e8kVcRixBesrBhyzr0KFj26Vt1udCPBJOrmruZbNR1dIltR5JOS86JNekBwbeP/ixDJadustH6inXZAq29G3cCJ06qWBw6lT4+efczpFjsgTmZf2Sg/TExOQp0x6Es7OaoSAlD4/kliF798Ly5UkvuOSHONvA2GB4uCD97Fn1OG4cBATA77+Dsy4O9HZ2KmQ9HB6i5mq/vQvOTnnwDQshhAORIF0IYZ+TuxrhPZXU/WhjEmMwGDJ3Uejqqi4ALaP/gqr1iU6MJiQ6hJZ/tOR61HUJ0sWj4+QKaJBoe+codU16vDGeyEhVk34/lm4fqQOd1DXnma1Jv3fv/mkeN2PGgDlFC+bUg00KJTIuEr1OT4l8JYhKjMJgMmT6fJweN7e0Nek6XXLz+UGD1KjrgKpJj75gk9ZsTtunPSsuXVItASpUUM+LFgU3fay6cQyw6w1YVRYuzYcLs9Uyj6JJA51qGa/86jL1eOfYg2cwE6Ki4OOPoU8fOHEiRzclhHhMSZAuxOMufC+cGAfh+7P2Pr0bmJOu1PwbgbMPYKc20BCD0Zi2/609Op2qTb90KXnZ/v0QlRBlk+5+QfqxY/DGG2qKIctI2kKkYTbBre1w71T6aSyDIybYfpHS9Ek3xJGYmHYwOHssAUrqwDIrQbrJBMOHq//LlVOjyT8pIiNhxw4oUUI1rd6+PXMtdfKqmMQYzkWcS/Ody4yIuAj8PPzwc/cD4E78HTQt66Opp1SgQNrBPUEFzpA8xSCgatJjLkJc8ohy9mriM2SMUeOeJAXOly9D5cqpWqyYEpKP1fhQiLmk+sKH7YD81eDFs9DuKLgFpL+d42NhT2/1/+YWEHstC5nMvPh4aNQIfvxRtQJo0yZHNiOEeMxJkC4eL6Fb4dRPcFduXQNwZZHqp/ffUNhQF27tUMvvHIN9A+H4t5B41/579e7q4gngqQ/BswSQNoCOToy2O9BQeixB+t2kze7YYX+d6dm9W00nNGcOTJgA77+fue2KJ4wxDrY9py7G1z4Nx8bYT+eaXz2mCtLtNXe3N3+0PZamvqkDlTStUDIY3X3mTPjf/9T/BgMsW3b/7T4uTp5UXQYGDlQj6Ddrljy94+No4fGFVJhUgS0Xt2T5vZY50vO757c+d3F5uJYHfn5q8DlzqrHY7A5EZzl+QrfCXTX9QZa2H3cT1teDHZ1gXVU48z/CwuxsK+VNYwvDHdXUvmxvcPYCnwpQIZ0vijEWjn+d/FwzgDFnWmz98Ye6kZwvH9SsKeMpCCEejATpIs/StFRN8o5/C1tawaGPYF21jGvPngSaBoeHgmaCQk3VyLiJ4RC+D9bXgXO/wpERcOhj++93cgfDXRXspGCvubuLS+YvRHx91cXf//2fqjE5dCidIF2nS/sHfP65Cn6eeQZefDHzNwfEA9DMELIRbm62Oz6BQzs3DUI3qYt3vTtEHrSfzr2weoy9arPY3sBxen3magj9kwagDg+3XZ6VmvQpU9Qo2wALF9ofRf5xdTppSuzGjZOX+fjkTl4ehdO3T9s8ZkVEXAR+7n54u3qj1+mJiIvA2Rni4u7/3vQUKKBmKEjdSinDIH3fO3BpLqCa2md6+yfHwb0TqgZc7wn3TpGQkDx+iZXeU00LmpIhqQVWgdop0qWa4N3i1t9q7Ika49Xzer+TU5fAq1erY/fvv+HgQXUz2UrTkgetEEKIDEiQLvKkBQvURau7O7RrB3H37sGxr9QPfWBf8CiW/APugKISojgXcY5EU+L9Ez+ou8dUM8Tq4+CZv6HdMfAoAad/UjUSQR9C+f7pT12TNJ8z8TdtFttr7u7tnfkg3TIy9uTJybVjma1Jj4tTc+Y2bw4bNqiLoc8+y9x2RRYZY1Ut9LY2sPUZVSudl1xbCt7loP0F6HgDijxjP50lSL9z1GZxnCGOL5t/yYbXN1C7aG3ijHG4umbue16okHq8csV2eWaD9NhYOHoU2rdXz597DiZOvP92H0r0Jbi9J20glAssn1vp0rmbj0flTMQZ9Do9Z8LPZPm9kfGRnA4/TdNZTdHQiIyPxM8Poh+iktgSjF9Num+laaoLguXcbcMlv3o0RmPpDx4QoEZlvy9NU33E/RvCSxehw2Xwb2i/ub6zpxrhHaDWT+BWSAXdAB5FkoJfc/q/Z7d3qubyZXqp56U6q5r3HHDkCDRoANXVRBH0Tmphz7WVsKoULHKHnd3z3o1PIcQjJUG6yHN27YIePdRgSo0bq+laEi+tUYFnk0VQbzo8/5+1ebYjmrx3MhUmVbCOZp4jwvepx7I91aNXafCrAdfXQLneUOtHqPsrVEunGbBHUtVdrG2kYS+gLlIE6xzS92OpEfznHwgOTlqHIdU6DdFpaxw0jWPHVF/drl2T+yvWro14SPHx8NVXaqCm6tVh3jzgwizV37N4ewj6IMeahj4QU2LGN+HMJrj9L5TpCe4B4OqXfjNYS5B+ban6vplUVXmcMY76xevzbOCzFM9XnHhjPH5+mRvEzcdHDZKYMkg3m9UNrUC/QLpV7ka1wtXSbe5+7JhK36xZ8rI0NYvZ6fAwWBMIGxvAytJpBtF71Cy1sJkZpO9xcPr2aRqWbMjpcNua9IMHVdPpAwfSf29EXAT3Eu6x8+pOzJqZiLgIihRRQfKD1qZb+p4fPqweT51SXZTstuaw1KSnULiw+u4bjaleuHMUtraBFcXh744QcVD9vpTro4JwN38o8xqurhCT+tDQe6im8ZoGvk+r5u+WIN3ZR7Wc+UsP29rZ36mE25CvUvLNZ3jwOeoyEB2tbm6k+V2KuQI7u0LiHTXN49XFDnFDzK50WrEJIR4tCdJFnvP77+pxxw71999/4J5wQt1ZL5R0Vevqp0Z7dVAHQtRV14EbGVx9Wdw9DpcWqKDbErTGh6ll19eqGk97Em6Di6/6HEzxYIhWTXqNUVDsheR0nvaqRwDPpInPLcF+khhDDJ80/ITfX/qdQL9AYhJjKFoUzp9P24fRHnu1MdGJ0QR4BVCraC0KexVOtyb93Dn1GBSU9rXbt+HXX2H8eDUYnci899+HUaNUUFSwIPz5J3B5ARSoC01XqJqrZmtyNhOR/8H53+HmFhVk26NpcPQrWFYAluRTfVlNdtqfJ0aoGrWkOc8zpHcHl3wQdRb+6QJ3/gPUwHHuzqpzubuzO3GGOIoXV9/B+7VW1elUbfqRI8lpd+5UNeftK7ZnYZeF9K7RW9Wk27kgvpnUeKXooziFhe+Hk99B4Wegzv/At1KaLi4PTDOroCS92s10WGKCJ6FVsNFs5FzEOZ4LfM4apJvN6kZ07drw5ptqDI6NG+2/3zLtWsrnRYqo/y+kGHQ9K5+lpbvG/Pnqcdo09VjC3n1vj7Qn9IAAdTP18mX13GyGe3cSVWuc8D1QsB5EHoCIpBN1/mo278+XT42ObsPZR9WkJ6QYtU6XdAmrpXO+SCkxQl0XWOx5G8L+VTf8zv+unh/96qHnWbeMtZKma8DFOeo4aLMX2u6FZ/eAk8tDbSs7aBps3gxjx8LcuXZujgghcs2TG6RfWQwbGqr/N7dQF2giT9ixA+rXVwOygArY3LQw8Cypru4uLYDj36i79A5q/4391ClWh/0hydFkdGI03+74lq0XtyYnPPABBFeBvW/DhnpwZSHcCFZN5va8Cf90ht1v2t+I4Y4KPkANErfEB06MVc89iqnB4w68r0Z+t8crKUg/N01dtBvuYDKbiTfG0+npTvSu2ZvS+UsTY1BBekJCchB982baudAtypVLuyw6MZp3ar/DgX4HGFh3YLpBumUwotT9U//9F8qUUVPezJxp25c1r/hx14/UmV6HK3ev3D9xNoqJgVmzVJPqfftUy5SpU4Hoc1CsXfKFsHuhnMvEoU/h/2qo8RG2tYHzMwAYv3M8bmPc2Hg+KUIJWQ/HRkHJrtDgD/U9ttdkNDGpM7hrfgj7B5bkh+BqadNZuCdFNdfU6GxGsxmj2WgbpBtVkB4VBTduqOSalv4AWQEBcO2aOl8lJKjPNNYQi6eLGh7e08Uz3ebulnEWHmau6Uy7PF8FL81WqtYGrTar1gcPvd6FsKo0LPWDZQEQus1+Ojs3KSwj6OfY1HMOFP1fvnMZg9lA2/JtuRF1g+jEaFatUgFyv36qBnv1asifXwW+GzbAl1+qfs7nz6vm7ilFxkVag3TLzcp799TYH5lVvrx63LIFevWCSZPUc7s16Z5p+yQULpz8flDjj8ScWQlxN9TNvqbL4YUz4FZQJXDxVd+XRR6wuSWlS8OZ1C3/vZN+OO4eT15mmTfdcA/yBUG+p9LfKc2UfC4DVZMdc1mN9r6vr7oGvDzP2q9evScT/cfjb8HRL9XAdwfeg+iLgJ3K58hDUKiJagUAULCOukGYGKkGu93bT42rE5eZfgLZw2SCjh3V+C7Tp6ubtb/+iu1+p/jfaDZyJ/4OBpM00xfiUXgyg/R7p+DfV5NrWUq8rE7yIk+4fl0FZDbMieCUNO/Rxdlw5HOI2Icjuh17m8t3L/Nm9TdtatI3X9jMiC0jGLMjqfl55BE4M1ENdNM1Gl66Ar5V1YWAfxPofAe63IGgdIY3d/ZKbg6YumpK56T6q5/5BS7Ns/9+F19VexF9TtVcxoUQk9SH3stFzYfk6eJJdGK09eLNMnfuDz+kP6Db00+nXRadGI23q7d13ekF6ZaRtVP3txwxQk3RdO6cGnQqKxekjmLe0XkcvnmYDec3PNLt7tmjyur115P7gZYupakaK/cA1SJjdQX1Z6/WGnWxt3y5qo0fMyaLn3/UWTg1Aap+DZ0joFMoBDQFYMP5Deh1ejZd2KTSXpoLvpWh/u9Qthc0W6G+58DYHWMZuXWkSqdPivJM8eri3HAXjBmc41N1jYlLugi1Bun65CAdYN069fh//5d+N4+nkuKFfv3UDZBLl9R4DpYg3cPZQ43vYKdbh2W6MUutXHbRNFizBj74QNXQTpqEmsqqQF0VMITvUwNsGR5yw7HX1E3EgBbQ9iDUnQo6ffqZShUQWM7vlpt+NmlDt8HpiermYar5uW1cnAvrasJSf9Xi4t5puL0b1teFRW6wtACcnfJQu5kdToefppBnIaoVrmbtl756tbpBM2GC6pf/4ouqNv3ll+H559V4Bf/8o163V5NuqfGeNEkdm6NHZ62PesWKyf/PnZvcQirlckiai90rVZCu97CW37ffqht/H38MrnEnVNeSQk1Uqy5zfPKNP8txaooHcyJlyqjm8paB6+LjId41qfnUxT/U98BsTA7SYy5B4ZaqNUiS8+dV4Pndd+o7r7kUsP1e65zVtHGX50PtyfDMdnjhNJR9QzVN3/4CLPaGxV6wr7/9D8psUIPVXpyrWgMYoskXr2ZQiYhIlTb+ZnLrtL3vwI7OEHkYNjZR48ToPSB8N9xMp8lEeuJvwckJ6ob7yR9UK7tM2rwZVq2Cr7+GixchJEQF7elZeGwhfuP8+OHfH7KWRyHEA3kyg/Sry9RI17V/Uc+D3rMdHVQ4lpik0WsiDoAxFmfnVKO6g2rqbvlxCvoweTAbB3TgxgEKehTkhYovcPnuZW7HqiuRdefW0TawLTsu71DzhofvUhe2Fd9XzdTjQ9VFTPR5VeOlGeDOEWuQcif+Dr8f+p3DNw+rDbn5qyZ+hiioP1M1MXROCl7ib4FfbSj9WvoZ1enAu6z6P6nvXIxRffCWQMPLxYuYxBhq1FDJfv1VPc6Ykf5q0wvSLYG/t6t3ukF6YKB6PJui4UtCghpF9+WXk5vSV6qU/vYdUVhMGIdCDvFG9TdyPkg3G9TFIcC909yJVAGSpYkroMpe75XUlUIHmlHdrCFtrZLBoGpi3nlHjWis18PatagaoYMfw+aWqrXSifH28xO+T603KOl7HnUGnFTz8n+u/EP/Ov3ZdDEpSI+5BPlrqPwdHqpqn6LOEhodyshtIxm3cxznIs6p7z6oYNG/EdT4PuPPxLeqzdM4s+pM6+HiYX2MM8RZv1fDhqkmwH36pL/KakkV96dPq8EOQdWkW9aZUU26JRg6lc0TVHz2GXTqpJoTt2mTdBwZY5LPC3vegi0t4fauh9vQzY3qho6l+XzR59T5J5MsNzj+/jt52a1bqBZFO7up4CrqrGpCbM+t7bC7F5R7E54/DNW/Vfn5u4MaLKzjDXjusPou5bIz4WdINCXy1sq3cHZy5kz4GWv/75TjEOzapQKqn39W0/EtX67+j4iLYPLzk0n4PIG3a75NRHyEtZXZgQNQqlSq0cUzoWBB+92SSpVSr1lUrYpqceWUYl7CAnWoVUv9e+kS1Kunvsd6Yzi4FlDH7p7e6ibJjaS7XXHXoFRXeOoTAMom/exYfk/GjoV7WiCgU0H6hvoq6HUtqJaF77bJ5x9/qBZ2W7aoGwy//QZGfQFVQZN0bKP3SJ7RoWRnNW3r5QXq93X3m6r/e7uj0OGKqsix5+YmVbPfbCVU/Qoa/I5PtZ4UL67GE7Dh5Jrc6ifsb9Vq5/pqNbJ9y01QeyI0WwWlX1U3IcL+gXMz4MLs9Geqib0BwZXV8eZbSd1QvZf56WctXSgGDVKPbm7JrSjsWXl6JYF+gaw8vTLT2xBCPLhcD9LnzZtHq1atqFq1Ki+//DL779OZdO/evbz88stUrVqV1q1bs2DBgqxvNP6Wat5omapjaQF1JzKzzEZ1B37nq7CrF+zuDbHX0DQ4fhzWr1d/V66ggptrK9QF6vGxatCux53ZkD3NCc1G9Rn/k/QDeX0NXPyDSpXUgDY2m/AqrWqGYy5Dseesgevs2dCoEXTooAYbGzXq4bOVnqNHVZPWsWPVxcWtW9htyrn/xn5c9C5M2TcFHToO3DiApmkEnw2mf53+FPMpxuaLm1Vtg5Ob6rd2eqKa5/xs0sTJzl5Jd+EbwY6OAEw/MJ1PNnxCv9X90DRNNf3TzHAzRdCn91Y1ZqGbVV/0/FVT74atVBfWliDdyzW5Jj3GEEODBmr3TJnoGujnB5UrJz8vXjwpSE9ap5dr+jXpVauq2t4lS5LL/9AhdSGWmfmrc5rBkGKwJrMBri6HIyPh0BA4Mznd9226sIkKBSvQo1oPNl3YhCm9PtkP6+4JWF1eNa8EODeVIj6q46hlJGcrj2Kqf7ZbQag3M91Vbtum/v73P/jiCxXAfv45sOMldROpwRxossxaO56GMQbQqe/lhVmwsTGc+JadV3fi5erFoLqDOBRySN3M0nuqGjhQgdr5GRB/k2Unl1GveD2eLfcsi48vVkGnR3EICVbHjyUITY9fDZuncW4qQknd3N3SjeL2bejfX9U8pSfloG8WKZu7e7h4pBuklymjunSsWJG8bOtWu0kzLSFB1az26KFqVl97DX75BTVmRUxS5+H6szJeSVwonPoZ9g9WLXquLE5nY7fVOcolHxz+THW1+W9opvP61FPg7Ky+U7t2qTESFsy8Ahd+V7XylT9Xs3iUfcP+Cm6sU78JQe+rICrqjJqCL+EWPP2Zuga4slD9ZuSy07dPczfhLvOOziPBlMDp26fJl091F0r5G2dpnZLye+XqqoL0Ql6FcNW7UtCzIBFxEZQsqeaXh+SuGVmh09nvMqTTqRp9ixo1UMdXwbrJC/0bERRkG8wDmFwKq8BaM6ubzB5FVWstvQeE/J9aj5O6JrNse9QoqFVLfV9xcgOfpLtXllZyzh7gWwXO/E/1L08a5HT4cHj2WfjrL/X/ihXgUriuGvzSpqY6qYmAzlkNkrmvnzo3hm1XM5+4+qlzUqqZH6xirwA6NSDdmUmwthIc/YJq1dT39kRSvLxgAWp/LQF0m6T8x91QFUb5KsKpH1WLj9MTYUcH2DcgqfWJBrd22N/+lb9Ua7lmK6DIs1CiE3gH2k9rh2U6ycz8diYYE1h3bh3fP/s9e6/vJSQqg5OfyDtirkLoVnXDKTr3z4dZcfw4DB6sru179YIhQ1QXszfegKZN1fKXX1atavKqXA3Sg4ODGTt2LAMGDGDFihXUrl2bvn37ciOdX5WrV6/Sr18/ateuzYoVK+jfvz/ffPMN69evz9qGvctC7GVISOrL5VkKezVE6ToxFo59DbV+hipfQODb3ArVUbWqatZlMKgfs//+Q41kem66GiG5+Ivqx0DTuBB5gZcWvMSgtYMwZ3FQndz277/qi//BB6o529dfQ+j5C2pfd7+l+jgfHan6PD+MiAPqR6heUrWs3gP07rRqpQbEGTdOjWA7YQJEeCTNVbTrdbg0Hwz32HM4gN691QG7ciUsXqz68uWE77+HunVVP9TXX1c3BoxG0jRjRdM4EHKAm9E3+f7f79FQz4+HHefqvausPr0avZOe4LPBqo+dKVb1Zas1QTXlc/FWn0PYDghoZm0NYjQbmbx3MtNenMbZiLPsvrZb1SC65Ie9feG/4RB1Dpz0qlbr7BR1UXC/GrOittNuRfuq6kFLrbelaXqhQvaDkvRYppYCaNs2883dPT3VyXfzZnXjpVs3+PFHqFIFNm1KbmL/KAe/0TQYOlTVGH35pZoqa/x41Od+/Gso8zqU7wteZdKdcmfDhQ1EJUQxYdcEIuMjORBygMuX4Ztv1Dp//ll9zy9fVsfcoEHqO/fDD7B7t91V2nf2f+DiAw1mq+fuRahbPRxvbzUvd0yMuuHx99+osr+yWNX2xKaO4JPZHYnblADhe6F0d9W1YkcH1e/cnnxBQFLNUdUvoURHQN24MJqN9F3dF0CN1ZC/KoRugYQIVUubZNGJRTxV8CkqFKjAohOL1MLiL6obpPv6q8eMFH8puTbQyZX4QuqGQsogPd4YT5Ei0KRJxquyqFfPduA3d3c16GLqPumanRuaTk6qibylJr1DB3W+exiapm6iuaQeq8q/kTrH3NwCBWqlvwJTIqyvDdFnVa1hhYHJLRZS8y6vAqJ7p6Da11AonRs06fDxUTX9ERHqXNqzJ+hMSQe1W0EVeP/9kvqzR++uxtEwm9QNgyNfJAdahjuqafX1VQ7R3P1MxBkal2zM+/Xfp1bRWpyJOEOrVmrfly5VaaKjk2vVU7au0DSNyLhI8rvnByC/e34i4yLR6VSLiYfRvXvy/3p98nfZMmp5YGCKgeT8G6lHnR6KPYdeDy+lKppEn3qq//Xlv1TTdBdfFZgXeVaVw8GPrAF0s2aqZY/ZnKrrTKrfI1z9oHALNQbFpsZwfRWaps5jaQZuK9xKnYt2J82HZoxSU5KCupFd/h0o2QVwUucCwx110EQcSP8Gk0cJQIOo0+oc4lYI4kJ44QX1+9+kibrhMGgQaj/vHFFBv2XwO7dC6jiJva4Gc9WMcO+4OufWGKdubEYcVC0N7DEngi7p5kbIetj+vLrezKQGDdSjpfsOpP/bufXSVuIMcdyIukEBjwKsPrM609vJKrM5eWaMo0chPDzHNpWrbt9WXTLGjlU3JH/5Be7ceYQZ2N1bVYIlRqhrk9v/PsKNP5x799TYVFFR6ibYnDkwcqS63v/3X3U9uHixanVkaZmTF+VqkD5r1iw6d+5M165dCQwMZMSIERQpUiTd2vG//vqLokWLMmLECAIDA+natSsvv/wyv1uG+86sMj1VjczfL6rnCarzUJwhjq+3f83o7aP592oGX1adEyqoN6s+cdtf5NLBvRw/roKPF19UFxjPtdVU80yf8mpQs9M/w/bnSTRE0X1JdyoWrMj2y9v5addPWct/Lhs6VDXhnDBB3aX+4gsonLhcjdpaZ7LqOnB7t7rozix7U354lVF9om/8n0oTcxHO/8bgwSoYHjYMihSBTz4Bo1spKN5BbXNXDzBGUaCgHjc3OHlSNY83m9VdNrvuHFOtKULWq6aSN7ck1fClYjapJrb7B6uLjaOj4NSPnDmj+ufVr69qMKpWtXORkGT/jf1Men4SVz+8yju132H/jf0Enw2mlG8pXPQuVCpUieCzwWhF2kK+p2F7O9WH7e7RpObv78GJb+Hf1+DSnwAsP7mcq/eu8tPun3DSOfHznp/VBVDFQerC6MRYdSJ284cKg9QJ+dDH6kLVxcd+RgGKd1S1BABFnyfGQ/U/TBloWKaR+v57dTEH6oaFfzrX76CCDlCf0Ysv2jZ3z6gmHdRNIRcXNZjS4sWqxm3IEHVXtW5dNSdtlSrpbzu73b6t9r1lSxVADx0Kn36Kai7p5q++x1cWwd/t7d640jSNDec30LR0U6oUqkLFghXZcH4DnTqpH5hRo9QNsY8/Vs04R4xQx92nn6rvfoMGZH7KnPzVVH/LyKS2mOem4RK1n2HD1MVwkSLq79tvURetaCoQ2vNmUq1O2vW2aKEu2D/5RN0MCw6GufPdIKC5Gh8iMVLNTRy+x36eCjVVLTb+6aJGWE6qJdt0YRMty7Skaamm1C1eV/VLrzhY1Rytq2oN0G7GhrP90nbORJzh4M2DHAk9ouabLj9AHQPnpqmLf9d0DkgAtwKqFgqgXG/idKpGL/Xo7qCCZeek4S8CA9MZTAuV5vsUrez79Us7cJyGRrzR/shzI0cm9flFtVRIE1xnkbu7ap4/d666kbR+fVLLotKvqZZlW1urQd4Ae+WMZlKBjXth1crp2Bg1cKVFyu9csRdUoL7teTg2Om3f8UxM6TRqVPL+A5i9KoJfTfhvmBpDJiCDu4KlX1X9j/f2SWqN5KryXbCeOndHHrYd6Tu18H1w6BN1c+fqMnXjN+4miaZEpu2fxuS9kzkfkT3VM6dvn2ZAnQH8/NzPvF71dU7fPs3rr6vfkVdfVee00qXB11e1QBo0SJ0DR4+GkV/HkWBKwM9d7Yufu5+1j/qIEWqGAVDN1OvWTS8H9nXsmDzPd7duWAejey4pTrYJwgu3Uo/F2llv3IwenXzjrkkTKFK9tepesKuHmubP0oS7wmBVu376J3XsuxXE3V2NbWFRtGjSYKFPfaxqnkF1USnRSb3fsgzQ+dehTx/12zB5Mmzfro7DBM1PTfVmtjR10oF/fRX47+mjWu2FblHfzXK94eR4VUngkc4BDirw9qkIO15WY7vEq0HfevdWze0jI1XA4OaGugHgUUx1KVmdNAheQDN1bfh3e/VbnHhHtUDxLAFXlqjrVJd8qom/PSU6qfPh/sEqHylumpk1Myazye5NQIsOHVTXmp49VeVCmzZJg4basfLUSsr5lWPb5W2U9C3JytMr2bNHdXP680/VdH7DBttpJy00TX0fX3tNtYJbs0b9VthreXfwoCrv2bPVc2fnB2sNkhc8/7z6vD/9VB3X776rjvNHQtNUC5aAllCkrbpG2Tcg+2b2SOnOMfVbcG66+jv6FYRuxWhUZXv2rBoo0ub63N5vRIprGycn1ZIoPj65ciYuTnXTuXs3+XsYEQGx6UyAlBc459aGExMTOX78OP369bNZ3rhxYw6lM+rQ4cOHaZyqDVbTpk1ZunQpBoMBl0xexZhd8qO1+hfT+d8AMJV7h/giHRmzbTQBXgGUL1Ce2Ydm46Q5UTmgMvfi7+Gid8FgMuDt5o1XhU/QuRaGQyPUhWvJ7tQuX4atW02sWKEupl1cVN+e3l23wvnf4OAnoHOHikNYe2YDDYo1oE3ZNjQr2Yw5R+Zw895NCnllbvTkI6FH8HD2oELBCmiaxtZLWymTvwxX712leenmgOr37O/pT8l8pblwQY22bTCoH21DwUPkd89PWb+yGM1Gtl/aTv0S9Ym46c2NG8kXoP6FjFww/02dYnXI55aPi5EXuRN/h3nzajJnjgoUChRQJ9perw+kaFUPODxC9RPzKq/+7JyFd+xQtQFVqqgLR7MZKkcb1UjGljOUZcSkZ/7FdH4O6MBkMkPgAPz9TRw4AD/9pJrotmunmlCbGsxT3QqiTkPx9pQr1Y69e03Mm6cGSdLrVUDx5vt2zoJnV0PcbcjnBNFnIHQ7OHmiFaiDltTKQodOXboWaKj6gmkuoPOEO6eZONFEnTpqrmnLhWX//kl9Sy2jQJlMxBpiqVG4Bm3LtaWoV1E6BXVi2cllxCXE8WXTL3mjxhtExEbQZ3UfrkeFUrT5JtVMOuo0lO4NgW+rH3DPshCyEXwqQ+XRnDi1k5kvzuSNGm9wNvwsw7cM517cPbwqjQKXAHXzotjzqlZX5wTNN6imwl5l1RzYlnLy8kpbZrWnqIvUqqNxv3ONjhU74oQTJpOJ6gHV0aHDZDJRqxZMnareO2uWyVpzl3L/LerWVQP6VK4Mzi5mni37LMW8i2EymSjmVYy25dqSaEhE76RP8/5GjVRTwp9+UjcCPv1UXUAWL65GRL53Ty2LSYgnJCoEnU5HoikRXzdfPMwBnDqlIyBAjZYM6lh1cU8kwZiAk84Js2bGw8UD58RIVYPs4gdoEHedCPdSHI28RNVCqpvAydsnqVCgArt2BbB4sTomPDxUjf8HA+aqz3jvIDWnb4X3uRxvJOzKHuoUV21G91zbg6veldpFavNLm1/w9/KnUsFKnI88z7x5JubPV7Xp+fOrmpk331TNSH/+WV3I6HTQsCHUtUxI7OtrPXauRFwkJCqEUr6luBF1g3xu+ahQpg845cd0dia49cNUojMUaMSQISYqVYKFC1Xe+/QBk3dFaL0raVAjdwjsBzin+X54eakL0RkzVPNSvV41MzM1WqYC5KNj1Heu6ljuxkRw6vYpivoU5W68ymflgMo4NVmtmqzePQYlXiW+dA9KXPiSsa3GUrFgReoWrcu6c+sweZSG1v/CmSkqYKw9laOJ7rxV/S2mtVfzRH28/mNOhJ4gMKg9tNyh8u9ZCp76KOPv+dPDwaUAVB2D272btK/QHledKyaTiacKPEX9YvUxmUzUr68uSoODVXm7uKT/PX/lFdXk1dcXOneGpctaUtK7JCaTCX93f9pXaE+8IR5XJ9c073/6adi2zUR8PPTta+Kjj+xf1J4IU01oKxVSN9O2XtpKjcI18PNIG4ROnKhq9rZtU+fhGjXA5OShyvnUD6rWuWhbbnlW59SFrTQrrQLhQyFJvxmtd6mawP++ACdPqPBxOp+pHlpsUS037p2Fkj2g3NsZf/6pltWqpZr4T5yo8jlgIJhM61Wt69kZKjB7+gv7H4pXeXhmt/rtvbgASvWEYp2g7NvqfHp1FbgWhnKDIDFWBYlOzupRM6tW0JozmJ1UsB+6jVi9P9+d2k6AdwClfEvxw84fGFBnAJUDKicVmyn50WxUN0KdnJNaU5m5HXuXY+EnaVGmBQBHQ4/ipHOidpHaVA+ojslkom7Ruuy7to98+Uzs3q0u4I8fV4FxkyYq4Jw9W53/fHyg42sxtA9rT2HPwphMJoIKBFGvaD1MJhMFC6rjedYsdQPN1TX972l6ZfLrr+r9X3+d/FKjRuoG0tspipNCrSHoMxUEJy0sWlQdJytWqNZAZg31u3PqR9Uao/y7ULavah3W7P/gwkx1UzrofTCZePttdW2wa5caS8HVFUwUg7qzIWQdVP4CNCfwClTvPzNZBb2BbzN2rIm6dVWLq61bk8dBMdWchMm1DJiSzk8FG0OD+XB2qpoOrkRXtQ/egeBTBUJ3qGu82r+m05dLD803q+/k7f1Q5Hko+yZubiZ27lSf2507atR0k5MntPoXTo2DxLsqwPZvoZad+QVu7YHindUNggofqWBm//vqJlPQZxl/zy/MhNNToMgLmIp1YsOpNey+tptiPsW4G3+XLpW7UMqnHOfOQVhSJb6LizrGduxQM6Ls369uLLz4YopNpfhOmM1mfmn7C88GPsvhm4f5bud36J0T8PV1tg44qtOBwWhk/7WjXI+6br0hWatILapXz4/ZrNI4OamytbdL5cur7lJ376oXT5wwWaf0A2x+4+wyRMG9k+DqD+jAeBeTsx/bb5+nbrG6+Lj5cOPeDS7duUSjUo3SvD00OpSz4WdpUlo1mdpzbQ9FvYsTfaMERqP6/dXr1T5cNx3C192Xcn7liE6MZu/1vTQtUhWX2Mvqxqc5MWnAwBJEa6Wsn7vls5o1S9UC//BD8uwN3btDfreb6samKR5cvNG8KxCrc8VgMuCkc0JDU5UZCdFcvnvZ2nWquE9x7t0sxJYtUKFC8jVOwYKqMsHZWV0r63TqxpF/i62qC9F/I8HZD8r1IyYhjjM3j+Pj5kNMYgzuzu4EFgjk6t2r3E24i2tSN+FyfuXYc30PlQqo351rd65xNeoqDUs2hKjzqiuH2aBuMrn6g199dbzqfSDhBMY71xg3TV0nvvCCigWiolK0PkvvNwLAZMLDQw1KOWeOutZ2cVE3M6dOVeedsWNVdw4fH9sb3inFx6ubA5auRYUL2x8z6fq961y9e5UqhVXtz4GQA1QpVIUEYwJ3E+7i5uyGyWyimE8xjoYepaRvSYrnK05EbARHw45Szq8cIfdCqFiwImbNzH+3/uOZwGfSbsgOnZbRbbYcFBoaSrNmzViwYAG1LKOMAFOnTmX58uV2m7C3bduWTp060b9/8kibBw8e5NVXX2XHjh0EBARkats7d+7E/ZHMbSOEEEIIIYQQQkBtS7+h+8i1mnQLXaomDZqmpVl2v/T2lmekYcOGaJqGyWTi6NGjVK1aFb2lnW5uS12TbIfRbKTWtFoMbzKcqQem8lrV1+hXux8ms4mp+6cSa4jlnTrvkM8tH4sXqz6rllprvR5e7BRL1SlV+bHtj4zZMYb36r1Hz+o9093egqMLOBF2gs5Pd6ZG0RoPtj8pZXFeIYcsJ2GXlNX9/bjrR/69+i9xxjjaV2xP/zrpTO+Tg6Sc8oZ0yylVbVLPZT0pX6A8q86sYlTzUbQPam9nbVkTFhNG9yWqY/LPz/1M1cL3GWTyCWcqWpSj69ZleEztuLyD/mv7U6NwDYIKBjGyxchHnEvxJJ/7TCbV0iwxUdUs2owj4mCyVE73q10H/rn8D68vf535L8+n86LOHBlwJN3Wq2bNzOLjizGajbxS5RWcndIPlTZf2MyfR/6kfIHyjGg24r77dV+meAjZoKYMdfZRVbz+jcAjc5WQj9qve3+loWtDTnCC12q8ploi3T2hun5oZtXVqFAj1fXMwVy/rgahtozV4u2txjx6WKtPryb4XDD1itejd43eD7WuXAvS/fz80Ov13LZMhJkkPDwc/3Q6s/r7+xNmaauTJCIiAmdnZ/Jb2nVkgpOlbU4SvV7vOCfrTExmqtfreb/h+wzfNpxEUyK9a/W27sN7Dd+zSdu9u+0AMIoP79R9h8+2fEaiKZEe1XtkuP+v13j9AXYkSVYmZ70PhyonkSEpq/R90PADToSfwF/nT/+6/VWT/lwi5ZQ3pCmnmJjkASCAb5/9lm6Lu1G1SFU6VuqYLdsskq8I297ali3reiIkdXzM6JhqUa4FQ5oM4dq9a3zW+DM59nLRk3ju0+szHivGEWVYTikr5yz9NNNpHNy8XHPKFSzHayteo0f1HhTJVyT9baLnteoZTE+bQpsKbWhToU2m0maK3gtKP+Soj4/QgHoDOHz4MK/VeC2pnPRQsHpuZytTSpVSf9mtY6WO2fY7nGtBuqurK5UrV2bnzp08++yz1uX//vsvrVu3tvueGjVqsDXVPDT//PMPVapUyXR/9MfFmzXetPZLtwxulBWD6w3Gy9WLmkVq4qJ/sj47IXKTi96FWR3uM9WVEBlJdSFaJn8Z9vbdm0uZecJZAgVLf0lf3wxvTr9V861HkCkhngBZ7K37W/vf2H9jP88GPnv/xEI4gFxt7t67d2+GDBlClSpVqFmzJgsXLiQkJITuSVW/EyZMIDQ0lPHjxwPQvXt35s2bx9ixY+nWrRuHDh1i6dKlTJgwITd3I1e46l0zbKJ+Pz5uPrxX/737JxRCCCGEfZZAwWRSbSez2J1LCPFoVChYgQoFK+R2NoTItFydgq1du3YMGzaMKVOm0KFDB/bv38/06dMpnjSnTVhYGCEhIdb0JUuWZPr06ezdu5cOHTowZcoURowYQdu2bXNrFx4rt27don///hgs8xnkcaGhoQwcODDT+5PX93/y5Mls2LDhvulyez8zm0/xeMrt79/DkuMsb8jtz/9hPY7l5+hl8jh+5rnhSSpnR9/XrMrqdbPIWbk2ursjMJlMHD58mBo1ajhM36R//vmH+fPnExsbS0xMDImJiQC4ubnh6emZ7gB5mqZlKb090dHR6PV6POzMVeDh4UGPHj0oVaoUkyZN4sKFC/j7+9O7d2/q16+f7jqvXr2abvpTp06xYMECzp07h6ZpVKhQgZo1a7Jhwwbr/ickJFjXpdfr8fb2xtk5uQFIeHh4mm3qdDrr/sfExKS7T+ntf4ECBejRoweNGzdmwYIFBAcHk5iYSL169Rg4cGCG68oo/fz581m0aBGurmpaJ6PRiKenJ0aj0abcPDw8iIqKwmg04ufnZx1DwVK+ZrPZuj13d3drOZtMJu7evYufn1+G5W6vnGNjY4mPV/M1u7i44O3tneE6DAYDsbGxGI1GdDqdNR8AkZGRNnkEVSYFkiaOz2w+c5rlO516Wscnwd9//83s2bMxm805fp5J/Tn/9NNPVKhQgRdffNGaJjuPs5R5HThwINevXyd//vzExcWRmJiIpmk4OTlZBx11dnbGy8vL+hsQFxdHfHy8zXfYzc0NLy+vR3qcxcbGEhdnO2ett7c3bm5uACQkJODq6kpcXBw+Pj60bduWrl27WtOGhoby2WefMWPGjFzrDuZov2cpP/+nn36aL7/88oG/Z4MGDbIZI8doNOLk5MTChQvR6/U253uLd999l6ZJIxPFxsYybdo0Dh48iNFopGLFirzzzjsUK1YMyLnyc+Qyud8xcefOHetUdxYpf1sSEhKIj4/HaDSi1+vtjlVkMBiIjo7GbDan+e1ylN+mB5H6PPuoyznl9rPjHD9//nxWr16N2WxOk37x4sVs2LCBe/fu4ebmRokSJYiIiCA+Pj7NdaOFvWuQ/PnzExsbm+2/Z6lldB2c1fQGg4EJEyZw7tw5bt26xfDhw2nQoIHN+1966SVcXV2t145FihThl19+Aexfd/fp04dSKTpnT5w4kbJly/LSSy+lm0cLR4yjHicSpDvYl+vjjz/m7NmzuZ0NuypUqEBUVBTNmzena9eu/Pfff3z//ff8/PPP1guLlIxGIwMHDkw3/YEDB4iJiaFWrVq4ubmxcOFCli1bhtEy53MuqlixIs899xzz58/n66+/xtfXl++//54CBQrw/vvv233Ppk2bMkw/f/58Ll68yIgRIxy6nJ8kFStW5IcffsjtbDxyH330EefOnXtk27N8zvfu3aNv37789ttv+Pj4APc/blLLbPq1a9cyZ86cNIHu46R06dL8/PPPhISEMGrUKLp06cJzzz1nfX348OG0adOGFi1a5Er+HPk85+npScOGDR/6e2YxevRozGYzX3zxhTVIt5zv7ZkxYwanT59m5MiReHh4MGPGDC5cuGBzPsqJ8nPkMhEPJ+XvWW6Uc8WKFRk5cmS2neO7d+9OgwYNmDBhgk36Gzdu4Ovri5eXF7Gxsbz99ttEZ+MgxZnZz/R+z1K733VwVtMbDAaCg4MpX748EyZMoF+/fnaD9J9//ply5cqlWb+96+4tW7YwY8YMaxx09OhRJk+ezLRp0+77WThiHPU4yfUp2IStl19+mXnz5nHjxg08PT1tak1iY2Px8/Oz+77IyMgspU8tISGBuLi4dEfJ9/DwoFatWqxZs4ZXXnkFZ2dn6tatS5UqVdi2bRuvvZZ2JMzjx48TFRWVbvrU8wS+/PLLLFq0iCJFinDr1i30ej06nQ4fHx+bO7seHh7Wee7Dw8Px9fUlKirK7v7nz5+fiIgIfH19bWrgM9p/Dw8POnXqxOrVq2nfvr21+8Xrr7/O0KFD6d+/v3U7KW3atCnT6dMr57i4OGJjY/H19bXe0U85G0HKco6JicFoNGI2m63lHBsbi8lkSvcHw1453717F1dXV+tdaqPRyN27dylQoIDdO8pRUVHo9Xpr7UNqiYmJ1tojs9lMZGQk+fLls6kNul8+HwVLOT+JOnXqxKxZs4iIiMjx80zKz/nQoUOUKFHCptyzctxkNv3t27dZsWIFXbt2Zc6cOeh0Ory8vHBzc+PevXvodDprSxWDwcC9e/coWLBgmv10cXGxHmfAIz3OLC1V3N3dbWpjU36u3bt3R6/XU6JECRo2bMjJkydtgvRq1aqxd+/eXAvSHen3LOXn7+HhQatWrViwYEG2nM8jIiI4ePAgvXtnfrqd0NBQ6tWrR76kebBatGjBtm3bbNLkRPk5apnA/Y+JlNL7bQGIj48nPj4+zfWM5bfL2dnZ7jEFjvHb9CBS/5496nK2bD+7zvEvvPACBQsWxMvLK0361MFtqVKlOHfunLV1Yuq8m81mu9cgcXFxaVonPezvWWr3uw7OanoXFxc6dOgApJ2pKjPSu+4OCwujSBE14v3TTz9NeHg4V69epWTJklnehshG2hPMaDRq+/fv14xGY25nxUZUVJTWvn177fr169Zl169f19q3b69FR0c/UPrFixdrX331Vbrb/P3337Uff/wxw3ytWLFCGzJkiM2yOXPmaN988022pN+zZ4/WvXt37c6dO1r79u21iRMnahMmTNCMRqO2fv16rX379lrv3r213377zfqe9u3baz179tTat2+vffLJJ9rJkyfT7P/gwYO18ePHZ3n/X3nlFe3QoUPW5waDQWvfvr124cIFu+u4X/p58+Zp3bp101577TXtnXfe0X777bc05TZixAitffv22sWLF7X27dtrUVFR1tdSlrPZbNY+/PBDbdasWTbl/MMPP2hdunTJ0f18/fXXtXnz5mnvv/++9vrrr2tffvmldR9SH1OLFi3SBg8enGYdO3fu1Pr3759uPkXOMhqN2o4dOxziPJPdx5mmadrXX3+tbdq0Sbt586bWvn17m3wfPHhQ++ijj7T27dtrERER2qRJk7Rx48al2c+uXbtq7du317p06aKtW7fukR9nlvNFt27dtH79+mlz5szREhIS7Ka1nA+WLVtms9wRjjNH+T3Lie+ZxaJFi7RBgwbZnPtSn+9Tl9/Bgwe1oUOHahEREVp8fHya76Gm5Vz55dUySSm93xZN07RNmzZp7733Xprlr7/+uvbnn39qffv21Xr06GHz22XhCMdMdsmr5fzKK69oBw4csB5P9tJv27ZN69atm/UcvX379nTzPnDgwDTb2Lx5c56/bu7Tp4+2a9euNMvbt2+v9erVS+vRo4c2YsQI63WxPZbr7tRx0ODBg7Vt27al+z4LR42jHhdSk+6ALM0zvb29rcu8kqZ3iYuLs/6flfRdunTJcJsxMTH37bdtb9teXl7pNifNSvpbt24xZcoU3nrrLWuNeb169fjpp584efKk9Q5oeHg4sUnz0QKMGTMGf39/+vfvT+3atfnyyy+ZPHmyzf57eHhQuXJl2rVrl6X9j4+Pt8m/s7Mzbm5u6e7v/dI3adKENm3aUKBAAa5evWqdtcBSbjt27LAO1mHpo5dSynKeO3cuCQkJPP/88yxbtsz6Wbdq1YrDhw/n6H5GRUWxadMmvvzyS4oVK8a8efMYM2YMkyZNskmnaZr1jnhqHh4ej7R5mkjLcpzl9nkmu4+zHTt2EBcXR+vWrQkNDbWms+S7TJky1vf37t2bYsWKMXr06DT7+dtvv5GQkMDGjRut/fUe5XHWpEkTWrduzdWrVylQoAATJ04kPj6evn37pkmb8nyQkiMcZ47ye5bd3zMLy3ku9Wef+nz/448/2pRfmTJl8PT05I033sDJySnN9xByrvzyaplYZPTbkhHLb1e3bt1o2bIlf/31l/W3y9JU1xGOmeySV8vZkt5SDvbSN2/enObNmxMWFsbGjRvx9fVNk3dLa78mTZqk2YalFjovXzenZ8yYMTz99NOYTCbWrVtnvS4uVKiQTbqU192pm6p7eHgQExOT6W2KnJGro7sL+ywHfcoDxPJ/eoO6ZSW9PalPAsePH6dbt27WP8u6UgbIoJoMpbeNzKa/ffs2n3/+Oe3atePZZ5+1vl66dGnefvttpkyZwq+//gpAlSpVbJoWVatWzdpcsHnz5pQoUYL9+/fb7L+9k9799h/UoGwp828ymUhISEh3f++XvlSpUvj7++Pk5ETp0qWtUw3GxMQQHR3N7NmzrcsszflTsqxnyZIl/PPPP3z11VfWQXQsr1mab+XkflqaipYuXRoXFxd69OjBjRs3uH79uk26Y8eOcfv2bbtNNe+XT5HzLM09c/M8A9l7nFmOo4EDB6Z5nyWv48aNs+77H3/8wXPPPcfw4cOtgw2l3M9ChQpRt25d6zSfj/I4s5wvdDodpUuXpmfPnuzYsSNNusWLF1vPB6nPG45wnDnC7xlk//ncwnKea968uc3y1Of71OU3btw4PD09mT9/PosXL07zPYScK7+8WiYWGf22ZMTy2xUQEJDub5cjHDPZJa+Wc1bSW87RlkqClHm33Eht2bJlmvdZBgbNq9fNGalWrRouLi64u7vTqVMn63VxSqmvu1PLzHWzyHkSpDsgb29v/P39uXjxonXZxYsX8ff3t3vQZDW9PeXKlePatWvW55UrV2bRokXWP1B3/q9cuWIzsNuFCxcoXbq03XVmJn14eDgjRoygZcuW1pNayv159tlnmTRpEu3bt6dgwYLcuHGDKlWqpLv/ln5slv13c3MjJCTE7gAaGe2/Jf8XLlywybuLi4vdwT4eJL1lFNGLFy9y8eJFIiMjGT9+PDqdjs8//xyAAQMG8O+//1r309PTky1btlhbEKQu5ytXrlC2bNkc38/MjHy7YcMGGjRoYL2JktL98ilynoeHBwULFszV8wxk73FmOY6GDRtGr169+PjjjwE1su+6deus6StUqIC/vz++vr60b9+esLAwrly5Ync/jUYjt2/ftvaPhEdznKVm75hbsmQJ69evt54PUnOE48wRfs8g587nGZ3nUkpdfhcuXOD555/H29sbFxeXNN9DyLnyy6tlYpHZzzy1MmXK3DeNIxwz2SWvlnOZMmVs8nC/9CnP0Snf93//93+4ublZ+1qndOvWLVxdXfPkdXNWpT732LvuTsloNGbqulnkPAnSHVTr1q1ZtGgRkZGRREZGsnjxYtq0aZNt6VOrWbMm165dy7CZV+XKlfH29mbRokUYDAb279/PsWPH7N6lzEz68PBwhg8fTtOmTXn11VfT7M/ChQs5fPgw4eHhbN++nfz581O4cGFq1aoFwOXLl62DhbRo0YIZM2Zw5coVAgMDrft/6tQpChQocN/BL+ztf+vWrVmzZg03btwgJiaG+fPn06xZM7sDnWQm/a5du7h37x4A165dY+7cuQQGBloHy5swYQIFCxakY8eOfPnllwB899131KlTB4ClS5eiaRq+vr64uLjYLeejR49St27dHN3Ptm3bsnnzZq5du4bRaOSvv/6iWLFi1kFhQN2R3rVrF88884zdddwvn+LRaNWqVa6fZ7LzOHvqqaeYOXMmEydOZOLEidbj6Pnnn+fw4cNERkYSGBjIypUradmyJSaTieDgYFxdXSlatCgA69ato3HjxixatIjTp08zc+ZM3N3dadu2rTUPj+I4S3m+uH79OnPnzqVRo0bW15cuXUpwcDBjxowhICDA7joc5ThzhN+z7D6fg5pSLL3znL3zfcryCwoKYv369daBylJ/DyFnyy8vlglk/JmbTCYSExMxGo1omkZiYqLNfM9t27Zly5Yt3L59O93fLkc5ZrJLXizn1q1bs3btWsLDw+2mX7duHXfu3AHg5s2b/PHHH1SrVo1nnnnGmvdr165x8uRJm2MupaNHj1KzZs08d90Maho2y3Silil8La0qU14XJyYmsnr1aq5evUrNmjWBjK+7LTJ73Sxy3hM9BZsjMxqN/Pbbb2zfvh1QTbn79u1r7TcyZcoUAGuzzvulX7RoESdOnGDUqFHpbjMz8z1euXKFyZMnc+HCBQoWLEjv3r1tpn/o1q0bX375JZUrV75v+gULFrBgwYI0TTS//PJLgoKCmDZtGhs3bsRsNuPi4kKzZs3o06cP3t7eTJkyhfDwcEJCQggLC8PFxQVXV1diY2PR6/XW/f/f//5H6dKlSUhIeKD9Tzm3Z926dRk4cKC1n5O9zzSj9D/88AOHDh0iMTGR/Pnz06RJE7p06cLcuXPTlNvt27fp27cvrVu3xsXFhYEDB/LSSy/h7OyMpmnWE3KhQoWYPn06er2eW7duMXjwYJ566qk0fRuzez+XLl3K6tWrSUhISDO/L0BwcDDLli1jxowZae7i3rp1i08//ZQZM2akO8KueDQc5TyTncdZSqGhofTt25e5c+eyYMECtm/fjtlsxtfX1xocFS9eHF9fXwoWLMjAgQP57rvvOHbsGDExMZjNZuv5ZPDgwY/0OLN3vujevbv1QtVyPkg5a0WlSpWs73ek4+xx/Z5ldJ67X/ndvn2b3377jWPHjlm/hz179qR69epAzpdfXi2TjD7zzZs3M3HiRJtlAQEB/Pbbb9bnGf12OdIxk13yajlnlP67777j+PHjxMfH4+PjQ+3atXn99dfx8vKy5t1gMKDX65k3b571HGnZ1y5duvDpp5/y66+/MmfOnDx13Qzw9ttvc+vWLZttvP/++7Ru3ZojR44wdepUwsLCcHV1pXTp0rz++utUqlTJ+rmmd91t2f4vv/xC6dKlraPIi1yUa0PWCYdz8+ZN7Z133tESExNzOyvZIjQ0VOvfv3+m9yev7/+kSZO0//u//7tvutzez8zmUzyecvv797DkOMsbcvvzf1iPY/k5epk8jp95bniSytnR9zWrsnrdLHKW1KQLIYQQQgghhBAOQvqkCyGEEEIIIYQQDkKCdCGEEEIIIYQQwkFIkC6EEEIIIYQQQjgICdKFEEIIIYQQQggHIUG6EEIIIYQQQgjhICRIF0IIIYQQQgghHIQE6UIIIYQQQgghhIOQIF0IIYQQQgghhHAQT2yQPm/ePFq1akXVqlV5+eWX2b9/f25n6Yk2bdo0OnfuTM2aNWnYsCEDBw7kwoULNmk0TWPSpEk0adKEatWq0bNnT86ePZtLORagyi0oKIhvvvnGukzKyXGEhobyySefUL9+fapXr06HDh04duyY9XUpq9xnNBr56aefaNWqFdWqVaN169ZMnjwZs9lsTSPllDv27dtH//79adKkCUFBQWzatMnm9cyUS2JiIl9//TX169enRo0a9O/fn5s3bz7K3XjsZVROBoOB77//nvbt21OjRg2aNGnCkCFDCA0NtVmHlFPOu9/xlNLIkSMJCgpi9uzZNsulnB6NzJTV+fPn6d+/P7Vr16ZmzZp069aNGzduWF+Xsnp4T2SQHhwczNixYxkwYAArVqygdu3a9O3b1+bLJR6tvXv30qNHDxYtWsSsWbMwmUz06dOH2NhYa5oZM2Ywa9YsRo4cyZIlS/D396d3795ER0fnYs6fXEeOHGHhwoUEBQXZLJdycgx3797l1VdfxcXFhRkzZrB27VqGDh1Kvnz5rGmkrHLfjBkz+Ouvvxg5ciTBwcF8+umnzJw5k7lz59qkkXJ69GJjYwkKCmLkyJF2X89MuXzzzTds3LiRn376ifnz5xMbG8s777yDyWR6VLvx2MuonOLj4zlx4gQDBgxg2bJlTJ48mUuXLjFgwACbdFJOOe9+x5PFpk2b+O+//wgICEjzmpTTo3G/srpy5QqvvfYa5cqVY+7cuaxatYqBAwfi5uZmTSNllQ20J1CXLl20kSNH2ix77rnntB9++CGXciRSCw8P1ypWrKjt3btX0zRNM5vNWuPGjbVp06ZZ0yQkJGi1a9fWFixYkFvZfGJFR0drbdq00Xbu3Km9/vrr2pgxYzRNk3JyJN9//7326quvpvu6lJVj6NevnzZs2DCbZYMHD9Y++eQTTdOknBxFxYoVtY0bN1qfZ6Zc7t27p1WuXFlbu3atNc3Nmze1p556Svv7778fXeafIKnLyZ7//vtPq1ixonb9+nVN06ScckN65XTz5k2tadOm2pkzZ7SWLVtqs2bNsr4m5ZQ77JXVBx98YP2NskfKKns8cTXpiYmJHD9+nCZNmtgsb9y4MYcOHcqlXInUoqKiAPD19QXg2rVrhIWF2ZSbq6srdevWlXLLBaNHj6Z58+Y0atTIZrmUk+PYsmULVapU4b333qNhw4Z07NiRRYsWWV+XsnIMtWvXZvfu3Vy8eBGAU6dOceDAAZo3bw5IOTmqzJTLsWPHMBgMNG7c2JqmcOHCVKhQQcouF0VHR6PT6aytiqScHIPZbObTTz+lT58+VKhQIc3rUk6OwWw2s23bNsqUKUOfPn1o2LAhXbt2tWkSL2WVPZ64ID0yMhKTyUTBggVtlvv7+xMWFpZLuRIpaZrG2LFjqV27NhUrVgSwlo29crt9+/Yjz+OTbO3atZw4cYKPP/44zWtSTo7j6tWrLFiwgDJlyjBz5ky6d+/OmDFjWLFiBSBl5Sj69u3LCy+8wPPPP0/lypXp2LEjb7zxBi+++CIg5eSoMlMut2/fxsXFxXqz2V4a8WglJCTwww8/8OKLL+Lt7Q1IOTmKGTNm4OzsTK9evey+LuXkGMLDw4mNjWXGjBk0bdqU33//nWeffZbBgwezd+9eQMoquzjndgZyi06ns3muaVqaZSJ3jB49mjNnzjB//vw0r9krN/HohISE8M033/D777/b9D1KTcop92maRpUqVfjoo48AqFSpEufOnWPBggV07NjRmk7KKncFBwezatUqJkyYQPny5Tl58iRjx44lICCATp06WdNJOTmmBykXKbvcYTAY+PDDD9E0jVGjRt03vZTTo3Ps2DHmzJnDsmXLsnwtLuX0aFkGNW3dujVvvvkmAE8//TQHDx7kr7/+ol69eum+V8oqa564mnQ/Pz/0en2aOznh4eH4+/vnUq6Exddff82WLVv4448/KFKkiHV5oUKFAKTcctnx48cJDw/n5ZdfplKlSlSqVIm9e/cyd+5cKlWqZC0LKafcV6hQIQIDA22WlStXzjpAphxTjmH8+PH069ePF154gaCgIGtN+rRp0wApJ0eVmXLx9/fHYDBw9+7ddNOIR8NgMPDBBx9w7do1fv/9d2stOkg5OYL9+/cTHh5Oy5YtrdcW169fZ9y4cbRq1QqQcnIUfn5+ODs7p7m+CAwMtF5fSFlljycuSHd1daVy5crs3LnTZvm///5LzZo1cylXQtM0Ro8ezYYNG/jjjz8oWbKkzeslSpSgUKFCNuWWmJjIvn37pNweoQYNGrB69WpWrFhh/atSpQrt27dnxYoVlCxZUsrJQdSqVcvaz9ni0qVLFC9eHJBjylHEx8enqTnS6/XWGgcpJ8eUmXKpUqUKLi4uNmlu3brF2bNnpeweIUuAfvnyZWbPno2fn5/N61JOua9Dhw6sWrXK5toiICCAPn368NtvvwFSTo7C1dWVqlWrZnh9IWWVPZ7I5u69e/dmyJAhVKlShZo1a7Jw4UJCQkLo3r17bmftifXVV1+xZs0apkyZgpeXl7W/n4+PD+7u7uh0Onr16sW0adMoU6YMpUuXZtq0abi7u1v7boqc5+3tbR0nwMLT05P8+fNbl0s5OYY33niDV199lalTp/L8889z5MgRFi1axOjRowHkmHIQLVu2ZOrUqRQrVsza3H3WrFl07twZkHLKTTExMVy5csX6/Nq1a5w8eRJfX1+KFSt233Lx8fGhc+fOjBs3Dj8/P3x9fRk3bhwVK1ZMM+imeHAZlVNAQADvvfceJ06cYNq0aZhMJuv1ha+vL66urlJOj8j9jqfUN09cXFzw9/enXLlygBxPj9L9yqpPnz58+OGH1K1bl/r167Njxw62bt3KnDlzACmr7KLTntAOAvPmzWPmzJncunWLihUrMmzYMOrWrZvb2XpipZ5r22Ls2LG8/PLLgKptnzx5MgsXLuTu3btUr16dkSNHpgkaxaPVs2dPnnrqKUaMGAFIOTmSrVu38uOPP3Lp0iVKlChB79696datm/V1KavcFx0dzcSJE9m0aRPh4eEEBATwwgsvMGjQIFxdXQEpp9yyZ88eu4NYderUie+++y5T5ZKQkMD48eNZs2YN8fHxNGzYkC+//JKiRYs+yl15rGVUToMHD6Z169Z23zdnzhzq168PSDk9Cvc7nlJr1aoVvXr1svZ7BimnRyUzZbVkyRKmT5/OzZs3KVu2LO+++y7PPPOMNa2U1cN7YoN0IYQQQgghhBDC0TxxfdKFEEIIIYQQQghHJUG6EEIIIYQQQgjhICRIF0IIIYQQQgghHIQE6UIIIYQQQgghhIOQIF0IIYQQQgghhHAQEqQLIYQQQgghhBAOQoJ0IYQQQgghhBDCQUiQLoQQQgghhBBCOAgJ0oUQQojHWM+ePfnmm29yOxvZaujQoQwcOND6/HHcRyGEEE8u59zOgBBCCOGowsLCmDZtGtu3b+fmzZv4+PhQunRpXnrpJTp27IiHh0duZ/G+Jk2ahLPzw/3cX758malTp7Jz504iIiIICAigRo0a9O7dm6pVq2ZTTh9cduyjvXVu2rSJlStXZut6hRBCiPuRIF0IIYSw4+rVq7z66qv4+Pjw4YcfEhQUhNFo5NKlSyxdupSAgABat26d29m8r/z58z/U+48ePcqbb75JhQoVGD16NOXKlSMmJobNmzczbtw4/vzzz+zJqB0GgwEXF5f7pnvYfRRCCCEciTR3F0IIIewYNWoUer2epUuX0q5dOwIDAwkKCqJt27ZMnz6dVq1aWdNGRUXxxRdf0LBhQ2rVqkWvXr04deqU9fVJkybRoUMHVqxYQatWrahduzYffvgh0dHR1jSJiYmMGTOGhg0bUrVqVV599VWOHDlifX3Pnj0EBQWxY8cOOnbsSLVq1ejVqxfh4eFs376d559/nlq1avHRRx8RFxdnfV/qpuCJiYmMHz+e5s2bU6VKFdq0acPixYvtfgaapjFs2DBKly7N/PnzadGiBaVKleLpp59m8ODBTJkyxZr29OnT9OrVi2rVqlG/fn2++OILYmJirK+bzWYmT55Ms2bNqFKlCh06dODvv/+2vn7t2jWCgoIIDg6mZ8+eVK1alVWrVmEymRg7dix16tShfv36jB8/Hk3TbPKZeh9btWrF1KlTGTZsGDVr1qRFixYsXLjQ5j3ff/89bdu2pXr16rRu3Zqff/4Zg8EAwLJly5g8eTKnTp0iKCiIoKAgli1blqmyFkIIIR6WBOlCCCFEKpGRkezcuZMePXrg6elpN41OpwNUINuvXz/CwsKYPn06y5Yto3LlyrzxxhvcuXPHmv7KlSts3ryZqVOnMm3aNPbt28eMGTOsr48fP57169fz3XffsXz5ckqXLs3bb79tsw6AyZMn88UXX/DXX39x8+ZNPvjgA+bMmcOECROYPn06O3fuZO7cuenu25AhQwgODubzzz9n3bp1fPXVV3h5edlNe/LkSc6ePctbb72Fk1PaS4Z8+fIBEBcXx9tvv42vry9Llizh559/5t9//+Xrr7+2pp0zZw6zZs3is88+Y9WqVTRp0oSBAwdy6dIlm3X+8MMP9OzZk+DgYJo0acLvv//O0qVL+eabb5g/fz53795l48aN6e6fxaxZs6hSpQorVqzgtddeY9SoUZw/f976upeXF2PHjmXt2rWMGDGCxYsXM3v2bADatWvHW2+9RYUKFfjnn3/4559/aNeuXabLWgghhHgYEqQLIYQQqVy5cgVN0yhbtqzN8vr161OzZk1q1qzJ999/D8Du3bs5c+YMv/zyC1WrVqVMmTJ89tln5MuXj/Xr11vfq2kaY8eOpWLFitSpU4eXXnqJXbt2ARAbG8tff/3FkCFDaN68OeXLl+frr7/Gzc2NJUuW2OThgw8+oHbt2lSqVIkuXbqwd+9eRo0aRaVKlahTpw5t27Zlz549dvfr4sWLrFu3jm+//ZZnn32WkiVL0rBhQ9q1a2c3vSWALleuXIaf1+rVq0lISGDcuHFUrFiRhg0bMnLkSFauXMnt27cBmDlzJn379uWFF16gXLlyfPrppzz11FP88ccfNut64403aNOmDSVLlqRw4cL88ccf9OvXj7Zt2xIYGMhXX32Fj49PhvkBaNasGT169KB06dL07dsXPz8/9u7da3194MCB1KpVixIlStCqVSt69+7NunXrAHB3d8fT0xO9Xk+hQoUoVKgQ7u7umS5rIYQQ4mFIn3QhhBAiHZbacoslS5ZgNpv55JNPSExMBOD48ePExsZSv359m7Tx8fFcuXLF+rx48eJ4e3tbnwcEBBAeHg6omwIGg4FatWpZX3dxcaFatWo2tb8AQUFB1v8LFiyIh4cHJUuWtC7z9/fn6NGjdvfn5MmT6PV66tatm6n9t0j9OaR2/vx5goKCbFod1KpVC7PZzMWLF3F3d+fWrVs2+2dJk7qpeJUqVaz/R0VFERYWRs2aNa3LnJ2dqVKlSpom76ml/Jx0Oh3+/v7Wzxvg//7v//jjjz+4cuUKsbGxGI1Gm/KxJ7NlLYQQQjwMCdKFEEKIVEqVKoVOp+PChQs2yy3BsLu7u3WZ2WymUKFCdpuYp6zxtTf6eOpAM3UwrGlammUp16PT6dKsV6fTYTab7e5XynxnRpkyZQAVhD/99NPpprOXz5T5sfd/eu9Lr3tBVtn7XCyf9+HDh/noo4949913adKkCT4+Pqxdu5ZZs2ZluM7MlrUQQgjxMKS5uxBCCJGKn58fjRs35s8//yQ2NjbDtJUrV+b27dvo9XpKly5t81egQIFMba9UqVK4uLhw4MAB6zKDwcCxY8cIDAx8qH1JqWLFipjNZvbt25ep9E8//TTly5fn999/txv437t3D4Dy5ctz6tQpm8/q4MGDODk5UaZMGby9vQkICLDZP4BDhw5luH8+Pj4UKlSIw4cPW5cZjUaOHz+eqfyn5+DBgxQrVowBAwZYm63fuHHDJo2Li0uafc6OshZCCCHuR4J0IYQQwo4vv/wSk8lE586dCQ4O5vz581y4cIGVK1dy4cIF9Ho9AI0aNaJGjRoMGjSIHTt2cO3aNQ4ePMhPP/2UbrPz1Dw9PXn11VcZP348f//9N+fOneOLL74gPj6eLl26ZNs+lShRgk6dOjF8+HA2bdrE1atX2bNnD8HBwXbT63Q6xo4dy6VLl+jRowfbt2/n6tWrnDp1il9//ZWBAwcC0L59e1xdXRk6dChnzpxh9+7dfP3113To0AF/f38A+vTpw4wZMwgODubChQv88MMPnDp1il69emWY5169ejFjxgw2btzI+fPn+eqrr6w3Bx5UqVKlCAkJYe3atVy5coU5c+awadMmmzTFixfn2rVrnDx5koiICBITE7OlrIUQQoj7kebuQgghhB2lSpVi+fLlTJs2jQkTJhAaGoqLiwvly5fnrbfe4rXXXgNUIDt9+nR+/vlnhg8fTmRkJP7+/tSpU8caoGbGJ598gqZpDBkyhJiYGKpUqcJvv/2Gr69vtu7XqFGj+PHHHxk1ahR37tyhWLFivPPOO+mmr1atGkuXLmXq1Kl8/vnnREZGEhAQQM2aNRk+fDgAHh4ezJw5k2+++YYuXbrg4eFBmzZtGDp0qHU9vXr1Ijo6mu+++46IiAgCAwOZMmWKtUl9et566y3CwsIYOnQoTk5OdO7cmWeffZaoqKgH/gyeeeYZ3njjDUaPHk1iYiItWrRgwIABTJ482Zqmbdu2bNy4kV69enHv3j3Gjh3Lyy+/nC1lLYQQQmREp91v5BUhhBBCCCGEEEI8EtLcXQghhBBCCCGEcBASpAshhBBCCCGEEA5CgnQhhBBCCCGEEMJBSJAuhBBCCCGEEEI4CAnShRBCCCGEEEIIByFBuhBCCCGEEEII4SAkSBdCCCGEEEIIIRyEBOlCCCGEEEIIIYSDkCBdCCGEEEIIIYRwEBKkCyGEEEIIIYQQDkKCdCGEEEIIIYQQwkH8P8oILfAmm/nJAAAAAElFTkSuQmCC", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "seqlets = recursive_seqlets(X_attr.sum(dim=1), threshold=0.1)\n", "\n", "plt.figure(figsize=(12, 2))\n", "ax = plt.subplot(111)\n", "plot_logo(X_attr[0], ax=ax, start=925, end=1100, annotations=seqlets, score_key='attribution')\n", "plt.ylim(-0.025, 0.25)\n", "\n", "plt.xlabel(\"Genomic Coordinate\")\n", "plt.ylabel(\"Attribution\")\n", "plt.title(\"DeepLIFT/SHAP of Marginalization\")\n", "plt.show()" ] }, { "cell_type": "markdown", "id": "17859ca2-e51e-49ca-b686-5f2014f109fa", "metadata": {}, "source": [ "Looks like the main seqlets are basically being discovered again but that some of the other small spans of modest attribution characters are also being picked up as seqlets. Depending on how clean your data is, this may be desirable, but also could be noise. Ultimately, seqlet calling in one pass like this is very challenging to do without any additional information." ] }, { "cell_type": "markdown", "id": "87b09493-27fb-4e36-b936-c441848f1c1e", "metadata": {}, "source": [ "##### TF-MoDISco Seqlets\n", "\n", "TF-MoDISco is an algorithm for identifying repeating seqlet patterns. The first step in this algorithm is to call seqlets. Accordingly, it uses an approach that is somewhat similar to the recursive algorithm in that it defines a background distribution for positive and negative seqlets separately, but it has several more steps that are meant to be helpful in the final goal of finding repeating patterns. For instance, this seqlet calling method will first smooth the attribution values, uses a wide fixed window, and fits isotonic regressions to estimate background probabilities instead of the binning strategy employed by the recursive algorithm. Although the seqlet calls from this algorithm are also done using a greedy procedure, the recursive seqlet property does not hold for the seqlets called here.\n", "\n", "We can run this seqlet calling method using the `tfmodisco_seqlets` function. This function takes in an attribution track -- which should not contain hypothetical attributions -- and returns a set of seqlets with positive attribution and a set of seqlets with negative attributions as pandas DataFrames with the same format as the other motif calling methods." ] }, { "cell_type": "code", "execution_count": 7, "id": "e3b1d0c6-b892-4733-b1b5-2c6ca0052c8e", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(19, 4)\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
example_idxstartendattribution
00102110620.788878
109419820.685255
2097510160.623512
30100010410.146656
408859260.128889
\n", "
" ], "text/plain": [ " example_idx start end attribution\n", "0 0 1021 1062 0.788878\n", "1 0 941 982 0.685255\n", "2 0 975 1016 0.623512\n", "3 0 1000 1041 0.146656\n", "4 0 885 926 0.128889" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from tangermeme.seqlet import tfmodisco_seqlets\n", "\n", "seqlets = tfmodisco_seqlets(X_attr.sum(dim=1))\n", "print(seqlets.shape)\n", "seqlets.head()" ] }, { "cell_type": "markdown", "id": "6445338c-3e47-49e1-b5eb-1a1989a9d34e", "metadata": {}, "source": [ "It looks like the attribution scores drop significantly after the third seqlet. This is likely because the first three seqlets correspond to the three known motifs that we inserted into the sequence and the ones after that are just false positives. Something to note about this approach to seqlet calling is that it assumes a minimum number of regions will be seqlets and so will adaptively change the thresholds internally to make sure that number of seqlets are called. This means that the method can overcall seqlets when examples do not have enough going on in them. Here, we can see that 19 seqlets are called, in comparison to just 5 when using the recursive seqlet caller. For TF-MoDISco, overcalling seqlets is not a big problem because the seqlets are more rigorously filtered in subsequent steps, but you may need to implement similar filtering (e.g., based on attribution scores) to remove these spuriously called seqlets from your downstream analyses." ] }, { "cell_type": "code", "execution_count": 8, "id": "6d0e3b68-c7c4-42bc-a20a-b5020a926924", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.figure(figsize=(12, 3))\n", "ax = plt.subplot(111)\n", "plot_logo(X_attr[0], ax=ax, start=925, end=1075, annotations=seqlets, score_key=\"attribution\")\n", "\n", "plt.xlabel(\"Genomic Coordinate\")\n", "plt.ylabel(\"Attribution\")\n", "plt.title(\"DeepLIFT/SHAP of Marginalization\")\n", "plt.show()" ] }, { "cell_type": "markdown", "id": "0d141ec6-b875-4864-b96a-53186fa6f489", "metadata": {}, "source": [ "Looks like all three of the motifs we added in are being underlined by the seqlet caller. Similar to the recursive seqlet caller, the TF-MoDISco seqlet caller will first identify core components of seqlets that cannot overlap and then subsequently expand the boundaries in a way that can overlap with other seqlets. The difference here is just that TF-MoDISco will expand out the flanks by default because that is useful for the subsequent clustering algorithm -- i.e., the seqlet calling algorithm is purpose-driven for the rest of the algorithm, rather than being general-purpose.\n", "\n", "You might notice that the boundaries seem a little bit... offset... with a longer span to the left of the seqlet than to the right. This is because, internally, the attribution at each position is replaced with the sum of the attributions starting at that position and moving to the right. This means that the chosen position for each seqlet is the left side of the seqlet. From this chosen position, a fixed-width span is chosen on either end as the seqlet boundaries.\n", "\n", "By default, the TF-MoDISco seqlet caller calls seqlets that are quite long. This is because the subsequent TF-MoDISco pattern discovery algorithm uses a similarity score based on alignment and so it is better for the seqlets to have longer context and make sure to capture everything than it is to miss something important in the seqlet. If you would like to change the width you can alter the `window_size` and `flank` arguments. The `window_size` argument defines how many positions to the right to sum over when replacing the position-specific attribution, and the `flank` argument defines the additional number of nucleotides to add to the seqlet for padding." ] }, { "cell_type": "code", "execution_count": 9, "id": "fc058a0d-c66d-483b-9cb2-3f831200d2fd", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "seqlets = tfmodisco_seqlets(X_attr.sum(dim=1), window_size=11)\n", "\n", "plt.figure(figsize=(12, 3))\n", "ax = plt.subplot(111)\n", "plot_logo(X_attr[0], ax=ax, start=925, end=1075, annotations=seqlets, score_key=\"attribution\")\n", "\n", "plt.xlabel(\"Genomic Coordinate\")\n", "plt.ylabel(\"Attribution\")\n", "plt.title(\"DeepLIFT/SHAP of Marginalization\")\n", "plt.show()" ] }, { "cell_type": "markdown", "id": "d3a0c989-a79d-477b-b883-70622f86cbf0", "metadata": {}, "source": [ "By decreasing the window size to 11 we get shorter seqlets but also we begin to get an overcalling of low-attribution seqlets. This is largely because, when a seqlet is called, a window surrounding the calling position is blanked out, preventing additional seqlets from being called at this position. When the window size is decreased, additional seqlets may be called that pick up on the flanking nucleotides that do not themselves have very high attribution scores but have relatively high attributions given a background sequence. However, the seqlets called for the known patterns are still several times stronger than any of the overcalled seqlets.\n", "\n", "Taken together, seqlet calling and plotting of these annotations can be a powerful tool for understanding what your model is picking up on when it makes predictions. As we will see in later notebooks, the automatic partitioning of the genome into units that drive predictions will be a valuable step for many downstream processes that can be useful in summarizing the features driving predictions globally." ] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.12.4" } }, "nbformat": 4, "nbformat_minor": 5 }