Skip to content

Instantly share code, notes, and snippets.

@zhang-ivy
Last active June 22, 2020 19:50
Show Gist options
  • Save zhang-ivy/70ced48c913300a7f4c3f5e5b9ba3961 to your computer and use it in GitHub Desktop.
Save zhang-ivy/70ced48c913300a7f4c3f5e5b9ba3961 to your computer and use it in GitHub Desktop.
From Joe Cunningham
Display the source blob
Display the rendered blob
Raw
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Set-up Code"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"import csv\n",
"import pickle\n",
"from collections import *\n",
"from itertools import *\n",
"\n",
"import numpy as np\n",
"import scipy.stats as stats\n",
"\n",
"import matplotlib.pyplot as plt\n",
"import seaborn as sns\n",
"\n",
"from tqdm.notebook import tqdm as tqdm"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"\"\"\"\n",
"Utilities necessary for SH2-phosphosite interaction prediction.\n",
"\n",
"Excerpted from hsm/predict/utils/dpi_prediction.py\n",
"\n",
"Note: _compute_interaction returns **both** the logit and likelihood (change to source code)\n",
"\"\"\"\n",
"\n",
"DomainPeptideWeights = namedtuple(\"DomainPeptideWeights\", [\"domain_weights\", \"peptide_weights\", \"interaction_weights\", \"bias\"])\n",
"\n",
"def _vectorize_sequence(sequence, amino_acid_ordering):\n",
" \"\"\"\n",
" Computes a one-hot embedding of an input sequence. \n",
" \n",
" Returns:\n",
" - list. Non-zero indices of one-hot embedding matrix of a sequence.\n",
" Non-flattened, this matrix has dimensions:\n",
" (sequence length, # of amino acids represented)\n",
" \"\"\"\n",
" aa_len = len(amino_acid_ordering)\n",
" \n",
" vectorized = list()\n",
" \n",
" sequence_indexed = [(sidx, saa) for sidx, saa in enumerate(sequence) if saa in amino_acid_ordering]\n",
" for sidx, saa in sequence_indexed:\n",
" idxed = sidx * aa_len + amino_acid_ordering[saa]\n",
" \n",
" vectorized.append(idxed)\n",
" \n",
" return vectorized\n",
"\n",
"def _vectorize_interaction(domain_sequence, peptide_sequence, amino_acid_ordering):\n",
" \"\"\"\n",
" Computes a one-hot embedding of the interaction between the domain- and peptidic-sequence.\n",
" \n",
" Returns: \n",
" - list. Non-zero indices for the interaction between domain and peptide sequences. \n",
" Non-flattened, this matrix has dimensions:\n",
" (domain sequence length, peptide sequence length, # of amino acids represented, # of amino acids represented)\n",
" \"\"\"\n",
"\n",
" aa_len = len(amino_acid_ordering)\n",
" domain_idx_offset = len(peptide_sequence) * aa_len * aa_len\n",
" peptide_idx_offset = aa_len * aa_len\n",
"\n",
" vectorized = list()\n",
"\n",
" domain_indexed = [(didx, daa) for didx, daa in enumerate(domain_sequence) if daa in amino_acid_ordering]\n",
" peptide_indexed = [(pidx, paa) for pidx, paa in enumerate(peptide_sequence) if paa in amino_acid_ordering]\n",
"\n",
" for (didx, daa), (pidx, paa) in product(domain_indexed, peptide_indexed):\n",
"\n",
" idxed = didx * domain_idx_offset + pidx * peptide_idx_offset + amino_acid_ordering[daa] * aa_len + amino_acid_ordering[paa]\n",
" vectorized.append(idxed)\n",
"\n",
" return vectorized\n",
"\n",
"def _compute_interaction(dseq, pseq, weights, amino_acid_ordering):\n",
" \"\"\"\n",
" Compute the likelihood of a given domain and peptide sequence given an input set of weights.\n",
" Assumes domain and peptide sequence are aligned.\n",
" args:\n",
" - dseq: domain sequence as string\n",
" - pseq: peptide sequence as string\n",
" - weights: input interaction weights\n",
" - amino_acid_ordering: ordering of amino acids\n",
" \n",
" Returns:\n",
" - float: likelihood of the given domain - peptide pair under the passed model\n",
" - float: the \"logit\" associated with that pair.\n",
" \"\"\"\n",
"\n",
" domain_contrib = sum([weights.domain_weights[didx] for didx in _vectorize_sequence(dseq, amino_acid_ordering)])\n",
" peptide_contrib = sum([weights.peptide_weights[pidx] for pidx in _vectorize_sequence(pseq, amino_acid_ordering)])\n",
" interact_contrib = sum([weights.interaction_weights[iidx] for iidx in _vectorize_interaction(dseq, pseq, amino_acid_ordering)])\n",
" \n",
" logit = domain_contrib + peptide_contrib + interact_contrib + weights.bias\n",
"\n",
" return float(1 / (1 + np.exp(-logit))), logit[0]"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
"def align_phosphosite(istr, before, after):\n",
" \"\"\"\n",
" Aligns a given phosphosite centering around a phosphorylated tyrosine.\n",
"\n",
" Args:\n",
" - istr: the input phosphosite \n",
" - before, after: the number of amino-acids required on each side of the phosphorylated tyrosine\n",
"\n",
" Returns:\n",
" - string: centered at p-Tyr closest to center, padded w/ '-' if not sufficient to account\n",
" for a length of 1 + before + after\n",
" \"\"\"\n",
" # TODO: This is incredibly lazy ...\n",
" istr = istr.center(before * 10 + after*10, '-')\n",
" ptyr_idxs = [idx for idx, aa in enumerate(istr) if aa == 'y']\n",
" if len(ptyr_idxs) == 0:\n",
" ptyr_idxs = [idx for idx, aa in enumerate(istr) if aa == 'Y']\n",
" ptyr_idx = min(ptyr_idxs, key=lambda x: abs(x-len(istr)//2))\n",
"\n",
" istr = istr[ptyr_idx-before:ptyr_idx + 1+ after]\n",
" assert len(istr) == (1+before+after)\n",
" return istr"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
"# Load Model\n",
"_model_weights = np.load(\"hsm_id_pretrained/SH2.npz\")\n",
"model = DomainPeptideWeights(\n",
" domain_weights = _model_weights['domain_weights'],\n",
" peptide_weights = _model_weights['peptide_weights'],\n",
" interaction_weights = _model_weights['interaction_weights'],\n",
" bias = _model_weights['bias']\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"['Domain UniProt ID', 'Domain Sequence', 'Peptidic Sequence', 'Bound', 'PubMed ID', 'Experimental Technique', 'Experimental Value', 'Measurement']\n"
]
}
],
"source": [
"# Load training data. \n",
"training_data = list(csv.reader(open(\"SH2_augmented.csv\", \"r\")))\n",
"header = training_data[0]\n",
"training_data = training_data[1:]\n",
"\n",
"alignment_mapping = pickle.load(open(\"SH2_alignment_mapping.p\", \"rb\"))\n",
"\n",
"_amino_acid_order = [\"A\", \"C\", \"D\", \"E\", \"F\", \"G\", \"H\", \"I\", \"K\", \"L\", \"M\", \"N\", \"P\", \"Q\", \"R\", \"S\", \"T\", \"V\", \"W\", \"Y\", \"y\"]\n",
"amino_acid_ordering = {aa:idx for idx, aa in enumerate(_amino_acid_order)}\n",
"\n",
"print(header)"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "c894d90e68644f24b8f4ba55e9c9a4ed",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"HBox(children=(FloatProgress(value=0.0, max=361278.0), HTML(value='')))"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
}
],
"source": [
"# Match predictions with measured K_d's.\n",
"sequence_predictions = defaultdict(dict)\n",
"measurments = defaultdict(dict)\n",
"\n",
"bound = set()\n",
"\n",
"# Note, we filter out unbound domain-peptide pairs. \n",
"# In the major paper supporting these data (PMID: 23358503), K_d's were not reported (which makes sense for non-bound peptides)\n",
"for row in filter(lambda t: int(t[3]) != 0, tqdm(training_data)):\n",
" domain_tuple = tuple(row[:2])\n",
" dseq = alignment_mapping[domain_tuple]\n",
" pseq = align_phosphosite(row[2], 7, 7)\n",
" \n",
" \n",
" likelihood, logit = _compute_interaction(dseq, pseq, model, amino_acid_ordering)\n",
" sequence_predictions[domain_tuple][pseq] = (likelihood, logit)\n",
" measurments[domain_tuple][pseq] = tuple(row[-2:])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Directly fit."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The most straightforward approach is to directly fit (*i.e.* linearly regress) the predictions onto the $\\log K_d$'s. Formally, this maps the energy terms in the underlying model linearly onto the $K_d$. Here, I show two models based on the likelihood model (*i.e.* HSM/ID) and the \n",
"\n",
"*n.b.* This is (I think) the most straightforward / interpretable way of mapping between the model and $K_d$. I am happy to go into the math a bit more formally and / or to detail the more sophisticated ways, but the results are roughly similar."
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Correlation (Pearson): logit - ⍴ = -0.195 (p = 0.000); likelihood - ⍴ = -0.241 (p = 0.000)\n",
"Correlation (Spearman): ⍴ = -0.178 (p = 0.000)\n"
]
},
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAlYAAAE/CAYAAACEto0QAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO29f7RmS1nf+X3e97znwjnHYHwbl/Kj3wO2zLKjKwZaR8YlITmGXIkXXKNDMI1BQ+jFITExccYB75qRlayeaMwPx1HHtHCjydtBDTIZQLMgP0QyTEAbEwzEAFe8fbmIgjpeTW4Q4db8cd7dd/fuXVVP1a69q/be389ate7t97zv3rX3rvrup5566ikxxoAQQgghhHRnkbsChBBCCCFTgYYVIYQQQkgiaFgRQgghhCSChhUhhBBCSCJoWBFCCCGEJIKGFSGEEEJIImhYESci8lUi8oHavx8Qka+OOM5rRGS7+//zIvKfRWS5+/fbReQvpqu1tQ7fLCL/T9/nIYQQEbksIm/LXQ8yPDSsyC3ajCZjzL8xxvw3Kc9jjHnQGHNkjPlMyuMSQkgssYNGG8aY68aY59WOb0TkQqrjk3KhYUUIIYQQkggaVsSJiDxXRB6y/O2LRORXReQbd/9+koj8lIh8Yvf5X7H87ng3eturfbwRkXeKyO+JyNtE5Fzt+y8QkfeLyO/spg2/qFGHt+/+9n4ReUHtb2sReZOI/K6I/DyAL+h6Pwgh80JEXi4i94vIb+/05Em1vz1PRD4gIg+LyA+JyM9VYQ310AMRecfuJ+/dhUH82QyXQgaChhWJQkSeCeCtAL7VGPN6EVkAeDOA9wJ4MoATAN8mIn9aecg/B+BbAHwugH0A/+PuPM8A8HoA3wbgiQB+BsCbRWRfRFa7c75t97tvBXBdRKqpyx8E8EkAnw/gL+wKIYSoEJE/CeBvAXgRznTkJoAf3/3tHIA3AHg1gDWADwD479qOY4x5zu5//+guDOIneq46yQgNKxLDVwF4E4A/b4x5y+6zLwPwRGPM3zDGfMoY82EAPwLgxcpj/kNjzAeNMf8VwE8C+NLd538WwE8bY/6FMeYPAPwdAI/HmYB9BYAjAN+9O+e/BvAWAN+4C4z/egD/qzHmvxhj3gfgx7peOCFkVlwGcJ8x5heNMb+PMyPq2SJyDOD5AN5vjHmjMebTAL4fwK9nqykphj3/Vwi5g1cA+DljzNtrn20APElEfqf22RLAv1Eesy5Ij+DMYAKAJ+FslAgAMMY8KiIfwZlX7NMAPmKMebT225u7vz0RZ+37I42/EUKIlicB+MXqH8aY/ywiv4UzjXkSavpijDG2sAkyL+ixIjG8AsB5Efn7tc8+AuBXjTGfXSufZYx5fsdz/RrOjDYAgIgIgKcC+Ojub0/dTUNWnN/97RM4M7ye2vgbIYRoaerPIc6m/T4K4GMAnlL7m9T/TeYLDSvSZCUij6sK2r2avwfgbgDPEZHv3n328wB+T0T+ZxF5vIgsReSLReTLOtbnJwH8GRE52cVUfTuA3wfw/wJ4N868W98hIisReS6AewD8+C6VwxsBvEZEDkTkIoCXdqwLIWTaNPXv9QC+RUS+VETuAvC/AXi3MeYBAD8N4EtE5Ot2C3H+EoDPcxz7NwA8vef6kwKgYUWa/AyA/1orr2n7kjHmdwD8KQBfIyJ/c2fIfC3OYqN+FcBvAngtgCd0qYwx5gMAXgLg/9gd8x4A9+xiqj61+/fX7P72QziL+/pPu5//ZZxNKf46gB8F8A+71IUQMnma+vdcAP8LgJ/CmYfqC7CLGzXG/CaA/wHA3wbwWwAuAriBs4FfG68B8GO7Fcwv6u0KSHbEGJO7DoQQQsio2YUkPATgsjHmZ3PXh+SDHitCCCEkAhH50yLy2btpwu8EIADelblaJDM0rAghhJA4ng3gV/BYmMLX7VLGkBnDqUBCCCGEkETQY0UIIYQQkggaVoQQQgghiSgi8/q5c+fM8fFx7moQQgbkPe95z28aY56Yux5doX4RMj9c+lWEYXV8fIwbN27krgYhZEBEZBJbDFG/CJkfLv3iVCAhhBBCSCJoWBFCCCGEJIKGFSGEEEJIImhYEUIIIYQkgoYVIYQQQkgiaFgRQgghhCQiuWElIl8kIj8sIm8QkdPUxyeEkD6hhhFCuqAyrETkPhH5uIi8r/H53SLyARG5X0ReBQDGmF82xrwCwIsAfGX6KgPXr1/H8fExFosFjo+Pcf36ddVnr3zlK+/4Tsw5z507h3PnzjmP01YfQlywzfRHaRoWwlzaRch1ur4be798Gj+15zC16ykKY4y3AHgOgGcCeF/tsyXOdvV+OoB9AO8FcHH3txcA+OcA/pzm+M961rOMlu12aw4ODgyAXst6vTbb7VZ9zur72+3WrNfrO/6+v79vTk9Pb/tb/TebzcaIiNlsNnd8r1mf5v2o/7Ze581mYwCY5XJpANz2d819bjtuLDHH8/2m7e/1z9brtVmv18muoU/a2tjBwUHSOqd+pl0BcMMo9CFF6VPDQvQrlCHaRZ/4+mNdp9rK6enpbcc6PDzsRetPT0+j3yuLxaJV7ysNr//dpvn159n2PmhquYh43w2+59K1XZWmJ0Pj0q8QYTpuiNKzAby19u9XA3h14zc/rTl2iDC5OmHusr+/b1arVXCn3NvbU313b2/vNmFyiUHVAW0d3tWh235X73ShRltMJ277TSUmtmtfrVZmf3/f+XwqkSpBDOrCZHtem80m2blKe0G7hKmP0peG9WlY2fpj1Xb7bsvaczQHlF2NlXrZbrfm5OQku74PUQ4PD50a5ipaA2u73XbWmxL1ZGhc+tVFlL4BwGtr//4mAD8A4LkAvh/APwDwlxzHuwLgBoAb58+fD7kYll2pj1q6HmO9XnuNwkpYbWJp61guY7hNDFwdP3XJJQZaz6uIJDmf6wWdC+Q3rKI1DJH6FYqrj/tebF0NL9vLs82D0mcZSgumUGIGrPWi1ZsS9WRoMKRhpT1evYSM+NjJ8pa2ac56WS6XrZ4tV9nf33dOo/ZdUouBb4pys9morzNV3Wwv6FSGWwwo1LAKPW4Oj5WvvYR4FGwGWMmzAyx6zdDqcNtvbZSoJ0ODXFOB2hIiTLkbLYu/VKPakGmAxWIRPI2aqmjFQOMBaHuh+aYoXfcxlTetxBEm8htWSTRs6BgrTVvWPm+XAZbCI86Sp9Tj2rS6GqI3JerJ0KAnw2oPwIcBPA2PBX7+Ee3x6oUeq+mVMT2n9XrtbXdaD0DXUf5yuewlZqbEmAjkN6ySaFifhpUxYVPj1YvNZRRpPKebzYYeq5GWGIO4PtOgbZOl6cnQoKthBeD1AD4G4A8APATgZbvPnw/ggzhbWXOv5liN494D4NqFCxdCLoaFpbXETiNqDCvtCK3LKL9vYSptFQ+GXRWYXMMQoV+xaNpVvf2kMIqGWoHNkrfE6k5pejI0SOGx6rNMZVUgy2OlbQlyn6WafrS9gFwjfs1UoC+mICSOoSpHR0cUpgL0p2vp22NljF33bB7OrkbRcrm8dRzfylWW8ZbFYjE73UkFpmRYbbfb7I2RpaxSubFtLx8Rcf5ds3Td99uYl5jGUzZlQMNKTczUy+npaSdjqMlcUh7Mqcxdg7qAKRlWuwtiGVnpa7RbvVx8BrdtxaEt0L5t6brtOy5Pletvc1pB0wZoWAVNp4R+t4vHqi0ImR6r6ZW5a1AXUKphhYgYBXqsxleql0Efx62MKt9LpO38R0dHXk9Ws+21vdh804RdVtD0FcdQQnwERm5YoWOMVZ8BwNppaV9y4nrbyK0jLOlLpUEpcp7l1pOhQamGVVUYY8XiKs1UBc2XT2ybqKYIu+Zk8RlOsTmF2pZKh+YiavuezXM3tBhi5IZVVWI9VimXrDefv7YPaLaIqdoGPVbTKqvV6pYWNDWwPhPg0pWS9GRoMCXDirlV5lcqAbB17i5twrfkvIkt+adr+x2NQFXH1kzftHnSNIab7/gxL/QuYOaGVaoki672l6psdttI5dYClrBiawfr9dqbW2+9Xjt1pTQ9GRpMybCix2qexZVnpWubsG2iqjFMmiM7wD76S9W2my9erefDd/yh4y0wc8MqlcfKdpw+jKuTkxMObkdeDg4OOu1wUbXP0vRkaDAlw4oxVvMtrmmw1Pl22lbLaF6EXV6WsXXTej58L8T1ej1onARmblilirHyJQNN2S9cuwgcHR1l2z2BZbhS6YpPTyoNmWrsFUo1rBAZ/Jm7YbHkK7a8K6kN7rbRlsaAiZ3eCdlCpGlYpfJYaeO5UoGRG1ZIkCA0xYvH9/yH8DDV4wlz7PfJkrZUMZ6uZ+3SExFRrbYeMyjVsKoK0y2whJTmNJ1GzENjRNq8N108Vr4tI0I8C00jLSTGKvRe9xkngZEbVlUZIo+VC9/z17StrsZQvU3SsCq/aIxtX7vy6UnKxRklAhpWLFMr9RGyKwgz1QpCW76r/f392wLrXZtPu0ZrIV4FbVB9G6EvvT7jJDBjwyr1FInreL6p8qpdalYIavpjbm1g6V7qz9PVTl1erVSLM0oFUzKs2HFZgMc6p8tQWC6X5vT09DZhsGWP9q2QAW6PGQDag94rA8y2NN02WgsJQF6v19EvYtsoNGRlZCowU8Oqz/xVrnNW/eDw8PBW2636iFZX22KsUu9RyNJ/8aXO0GpMTOJkeqwKFCZ23OHLarUafO8/X6mMHN/3tJ4gzfXV9wX0LTMOHa3ZBOr09DR5nhhtygjGWPVjWKV84dgMfduLse057+/vq/pJNZ3t8mJwxeB4im/Bj7b/29pDDk0ZEkzJsMrdGOdQTk5ObnUUTb4TTUktuK4RUV/l8PBQdU5XkkbXy9MmUEON/IZewYOZGlZ95q+ql9Vqdccz7NpnfO2CA99xlLr33bXBdleN4arADAURq2pyN8g5lHqgdQqhdE3BxZRqVVzJo+M2g9Q1WovxAow9VgEjN6wQuSowlaGs6ZvNY6boM80A5qoe1cu55H7J0p5V3fbdsWtMn6BUw6oq9FiVV1KIY33qIFWeqb7iOQ4PD5PnwvJljK+IXdU19lgFjNywqkquGCttH60b7KlW7G12CzVoRI2vVDF1Mbs8kMfAlAwr7lc1rrJer5OJeTNupE3YY4W+mT09VdEIk+2cVcqHtuuaQqwCZmpYGdNv/ipXf3Al+AwtNKrGWTabjartTEFj+gRTMqy4X9U8y8nJiTHm9qmHNqPq5OQk2PN0eHh4q3350jeEFo0rPfQFVa3kGjuYsWGVglhPcD1HW9tG3/VS2qIVlu5FRLyas1gsJqExfYIpGVa7C2KZWbHlkWqWZlCmZvqjnnC0b49Vm6ci5py+0eQYgkZBwyqattim5qpAW9tpSzBrW3naJbcVS5lFO+vT1JgxaMqQYEqGFfNYzbdoBKHNQ6QZ2VcGUMrpjTZhsqVUiPE8LJdLdUJIbbzXkICGVRTaGC2Xwd7WBtpenJzum3dxxclWbaMUPRkalGpYIWJVDZfzsrhK2+bJxpy9NFyGWWWQpWpfbVvYuILQQ71szRIT1N8l0WgKMHLDCgn2CoxBu5hBm3E95lws8ymuBML1kltPhgalGlZVCRnxcQTVvVSejiney+Y+gtq2U72UNDFWmgDg2E2cK2LiZ2K8bjkDVDFyw6oqQ3usQtuRz3PVhiuWkYXFVuYU8I4pGVYcQaVr+CVOq6YQcdvLwtZ2ROSOqbT6CO3w8PCOabSYF5bG01D3XFXnDLl3MX0k15Jq0LCKIib9hqsdNUmZHoVlfmUuKRowJcOqRGOg9FIFtbbNheeum61jdknRELJtjIjctvolNEAzJCeR77u27UZcq7aa9812nJj71TegYRWF7fm6pmK0gwrXd1nGUTab9s3gQ7SkWUL0eC5JRUHDisX2sk+5nNo2Kg7NPaaZpnTlYnGNmFyGU2zixhBjzPVd2/XUl8fbRLMtUF573+mxGpdhZYx7JZ/NqLf1qebz59Tf+EpbnrvmBvTNbOuhqwO1xhU9VgbZRckEChNHU/GlrcH3fc5qv7KQ51ZPjGkr1Ui7SwZrbcD4UEIREzvjMuY0nqu27S00nroUgIZVJ3wDi+Yz9fWnymhPlYSZObD6L64NtH265Ys51Wyq3Cxt2x2VtBI5JZiSYTWX0VQf19n2gh7CUK2mKLTTU9X3tcHmMZ03pD5DubZjPHA+mvfHNorNsQs9aFh1wmWIu5bHs0yj+JK7+nQrZiCp0ZPqezk0ZUgwJcNqLh6rlNm/q7JYLO54oababsZXjLm9U/pGxZvNxly8eLH1b3t7e507aEg7GspjZdtVoO8MyH0YdBpAw6oTrudm+xuNq2mUk5MTr4b5DKS2d0w1w9Bn25wKKNWwQkQeGMZY5S2xWwo16SLw9S1oYtGef8hRVi4xCpmCTAlGblghUx6rCpdXQDPNk1tLWOKLb8WwT7dc8ZwpyKUpQ4JSDauqcEub8ZQYL1dbZ+3qeeyKS1hyxQXkEiN6rIbVr5TYpsI1z5Qb2o+72J5xW3LiJn1rDT1WIxSm3A16zqWKf9Iu27Ul7OyaK6crfcQAdA3WzCVGjLEar2FlQ/NMNf1stVoxCL3QonnGXQzvvtvf2HHpV3ZRMjSsokuuEefp6eltHdb3fZuREbP0F3BvWxNi2KRctZJCSLoco+u1cFXgtAwrY9qfqabPNfegdPXx+t+qGM7cujiHUt3r9Xpt3QPUpSdaremiC1wVODJh4o7r+Us9GaFmWs9nILR19NVqZfb29m77rM0Ddnp62iroMakE6svNRcQcHR2pfpdqBBgjRmMdHbqEaUylVMOqSehy+Qpf/24usR9qQQwLbunUxYsXb9PAo6Mj76o/n9aMVVeGwqVf2UXJRAjTFDtumxExVKeM/a1r9OPq0DZ8I+y2zu8Lpm+bxrCJg+9YLlHJGaypNepKG0GChtWguGJy2pbLV21lvV57VylXv623r5Q5seZaKl1JeR+1muTTldL0ZGhc+pVdlEyEMOVu7KnLxpGcsu9S5SHpUndjzjqZLxajDyMjVnDaAjw1x7IZh7Z7OEQwvMaoc+U1qq57aHHEzAyr3C8irfEfG//Y5t3IoWlTKxWpjCtNcLuvvbj0xDU9OSVc+pVdlEygYTW1dAua5dF9n//09DQ6b1ZdlH3f1U6LhbyAUtz7kGOF7EPYtjdXalf6dmvfuqZ+v7XG85CZ2DEjw6qEaRWtZzNmoEXPVH+l6ocpj6lpe66pxJA2kjp2qxQwJcOqi3el1BLaUFOXLqJYF2XfdzWJLkNfQF0FvV7/kL31fMH4Li9kaNxVyL2y3bMQYV6v14MYAZiRYZUqBq8L2r4V+hKnZ6rccnR0pBp4tbWVtpXfVYxraBupn6uEQUYKMCXDaoqrTkQkOvFm7rLdbtWxFJrkc6FxIEdHR9Z7qr33FSHPQCMEuXLFtLn6UxjuqY0AzMiwKiVhostTUP0ttE3E/I5lmBK75Y0rtMH1d825ShhkpAClGlaIyFw8xQ58dHQ02lHfyclJ0PdjjZF6qaYvbffs8PBQPb3Z7MxtqwK1v9W2Vd9IUesiD920uWsbS20EYOSGFQL0q9SXSd0oSu2FYCmj2Aa9rvhPn7aEPu96WyllkNEVlGpYVWXOMVZVg8pdh6FKrDHSLD4PmcaDpnU/xwpBqMs79PuhL2uNZ+Hg4KD3KcwKjNywqspYYqw0dQopbXEzufWFxd6v6//2xX9qtEXrqWy281IHGaFgSobV7oJYRlxc9D3ybZtS9NFFCGweqLbPtedxeRkqb17dOKrnHDPG7RUMSR7YFczIsDKmvIBd7SCmzfNbj5esXxeD2PMVW7+u60uVPsO2grvyYrmOV32var++fSmnmh8LNKxY+uq0oWW5XHqfry3hZ/M4sdehxWfAuDxPrheoTVh8dbYlX6zqttnlDnIFnRqjMxTrU6LL5VK18CAUzMywKo2QXRNccVmcAiyn2AZadaMqZAW4q41UmhLjOXcN/MYCpmZYTTFBaIpydHTUSwyaLabp4OAgOMaq6qyuEbvvGkTEnJycRKWI0LqbXXlaXHW3vWg0mepdq3d8L7Dqulz3rvqOb8RIj9U8DCvNtLBvmbxGi2NTubCElaYRVOmkdl9XW3ENYm3aZNOLqXirjJmgYTWHufyYLOyVZyO1ceWavjLGRGdYtnUobQB76NZGVWI717VUxE7/ue69JmeZTXQ0xqbv3jUThtquf6gYCNCwyopv8FD3dFT/jvFOrdfrzi93lrjSd/xu3ZuumeaeSnyVMTSsRltCjZUqfiilkGkyh9c7Vaix09ahUhiGtlGy7QWhzfvkm0r0CdnGkbPMNe2iOa7v3mnFa6hVO6BhlR1XDKDN6EpZOPtQfvF5rEKYyopAYyZoWPUx3VVqiXWjL5dL7xYzvrK3t+fNHN4mwKHnbdI1bqNuoFSdttmJNb+zfceY9hik7daeBb1e2l4mPne4xhNW3bs2o9q2jUWXIPqugIZVsQylsQx2z19cK4GBs4UKbZpej9usU4I3fAgwNcMqd0McqlTela4GUkyp9nvydQJtR7F9rz49Vyd2pZF2aa/r975j2xKJaurZZtRpgjddyUubgeVt320TQVu8g8ajlwLQsFLT14rCWA9pqJZo+wLLsKVKTr3dtu/1ure3Z80J2LagpZT4zSHAlAyruUwDVg3X57npw+jSbKbs8+y0bezq8hT5nrnrPrhSKISItyZIM/SYzXvSx/VrjdiQ7w2RGgA0rFRoFkSkOq4vpi+m7Q81rcgSV6r+bft7yJY4Gu0pLe1ILJiSYTWXOfmShadtLzlbZ6z+G2KEteHq+DFbMzSLJt1BRew9i71+3zU0f6+NY8gd7wAaViq008Cpjlu97GzezBAvcNX/64O1uWj4WEpl4MT8rkluTRkSTMmwyt0I515iAtTrv237XOshiZmf18ZrhcQWxVz7crl0Tq26rl8buB56n2zfi0miGgNoWKkIff5dj9tc6VW1iXpb7TLw88X0sAxbYnexCPFYaRZBjQ3QsJpfKTUo1JbATjPvHjs/rwlItyXfbDu+a/9AV7nrrruiYppCPRaa+2S7Xs3xUwEaVipCPZZdj7tYLMx6vbYmlHQZRpVR7tOf2MFZbIlJXzOn4jKyQ/S5rV34ts8ZK5iSYcWRjr6TNKfjSinNrLshe9N1mZ8PCdau6mYLrI8V6phVeDExNq77ZDtebAxYLKBhpSI0xi7VcX192PWyLS2UobT6dCmbDjnFQu5RZVRVbcWluy6NGmrv0aHBlAyrOQWvpyiV4VCKcaWJz6p37NSkSi+g9fg0SxuauISUAZ+h8RR91QM0rNTY2ptrNwDNs3KtNvW1iZhl9X2UKRlNMX0x5fFj+7ZLQ/uKu8odBI9SDSsA9wC4duHCBfXFlGRYDeU9i906pl7PmIShrrig5ve01xFyz4Ya0XTp+CHXs1gsWo8RY9h1IfRF5PKcdXHpY+SGFSL0qyuu/Gyhy9q7eKx8bbNrLjpfaTMo+zpXSaXtvqcyLLvojUtD+9C3EtI2oFTDqiohI74hR0KllFjvSNdSuYJ93zPGLqTVaqDQoNchO0mXjh+ylNy2kfHQIuEKMI2J9YoVSIzcsKpKjgShvmeheVaxWqptm314VFztbmh9HLrY7rvtHru0tplSp6veuNpbH/o29GC0DUzJsJqT27feUHKcWzNtt16vjTH6HctdL/Vcbt2uHb/ukrZ5B09OTtTH6Pv6Xdcbszox1qXvEqYxlRyGle9ZaJ6VVktXq9WtYPaYttmXcdWcjsyRSHmoYts5wRh37GjbgHyxWNyxD2RXvdEkBk15vhLSOrj0K7somUBhmpPHqp6VvMsxYn6nnbarXsZdVo244kQ0pOi0mmP0lRKij+vxHav++dHR0W0LHmyeNdu1uUTfhUuYxlSm5rGqrwpsa3+aQOb637VhDE960pOya26pxRdE3hzcVc/PlyohVGt8i2KqNtXMW5Z6oJjSYxWrty79yi5KJlCYtlvdfmxTKRUxU4GxRmj1onQZZfUVI5pGro25CEmf4FoOnroju0aF9U7pul++a9lYVvvYjFSNIeg7lm2qV7tdRZd7DhpW0Wg8BDExVr7nGHPeuc0y9FFs+QO7xM3Z0svYvFnaNtW2/c1qtUqqyammF7scx6Vf2UXJRAhTSKzOmEvTMAkJQD88PDTGxHn4KiNA66HQuGVD6tE2LegK2vXduxR09ZQ275nPe+drCxqB0yT2s31nsVjcYfD57kHoPQcNq06Eeo5ivbB1Yj1lLP0VjRcypNgWRXTZO7b5PW2703jIunj2u3i+MDXDandRky+azXVdxTZNp+2sWu+JpnF2MYRXq1XrKMhVUs+1p3ie9Tim0GeiMVLr7n3tFj0p22voPQcNq6QMEacXG9vF0l/pqz/Xi3ZLLt8sR9VO+0wInbI9u8DUDKs5eqyMCR+N1F+0leCGTKO63MLN5+HrAEOPZNtGHLEvnpj25vIWxdwLjZHq8zS2HSvltDo9VvkY4iVkTLzHag56nbNUA+Ec9znUY6X1EnXxJqVqzy4wNcNqLu7mptUc2mnarO7YHEY+NNMSfea1qZfUo5+Y9uYyfnzPwJWfyFWfajuRkHsTmyBSc899gIZVMoZ4CRkTH9vVXKFGQ+v2UoU/dDlGinsac4zQGCutl2iIlX+Msaoxl07ZjMsJ7XhVKoQ6occIacS+FSt93SfNcvAuL57Q9uYSSZ/HSuMljDVSbav36pn5l8uleh+3rps1g4ZVMoZcfh4T25VqYLVarSar/7kdBuv1OnirnLZ3TDMeq7kqsCSPlaY923DpV3ZRMhHClLsB2kofHd43Goxp9G0jCtvLNMRjFbIPn++atfda2xG6vHhs7a1tReL+/r71ReLKLVMdL2R6MmR6NzQvl+8ZpJhmAg2rZAz1EoolhWaLiDk5OYnaRaL0ktNYbOqOdiAcu9KvpBirLrj0K7somQhhSjn6Sd1I+0gFURfH+gu17cXeLG33rnndq9VKHahuwzU9FXq9vo4dkzepy4vH1cF9q1bqRtTh4WHr8+qa58UnyjH3q3ldqRMKGkPDKiUjeAllK6UbYjFGlU1Xm8eq/n10dNT6/Wog2IavXm0Ddy0pVv/WCkMAACAASURBVAXmxqVf2UXJRArTdrtVT1mkbtS+0seu4/XrrhtWvnvQzPjtG4lUHTa0Eae8Xt/S/pjpja4vntDpD5dnqu2edxEM3zMdMhtxCC5hGlMpwbAyZpiXUMw5SlhslPv8fRRNXr3T01OrUVUVm4E0Vl0ZCpd+ZRcl08GwKrGzVGKTcrqy8qrEeurqaO5ZW04kn1GR8h766rhYLKJeHn29eFJ4ULt4F3znL2U6qIlLmMZUSjGs+iZ2cDJE6IZm4caUSuXl9umyNk2NdrX3GHRlKFz6lV2UTKQwpeysj3vc45IcpzliSDEtKPLYtjax11xH60VxGXOhqRT6zMRcwnRHqrbYFKoQL5nNe2m7PyW42F3CNKYyF8Mqdjq9b4OnPp3l0jef52ZMRePlDtGl5kKUunOgbQ/G2N0gpoRLv7KLkokUpqFGJ9XIwPWd6sXWV8ftes11tIaVLydSXUxd9ap78JodLmabHl9dcnTuVG2xOeUbu6zdd/2lxOOAhtWoiF0A0rfHqh5E7Utkqd23cCzFpXGxutSWkHl/f9+7f2QJmjIkLv3KLkomUphSvZSrDhfzN+CxTWv7ytFUNxpiBKoZY6XtbL68KppM4EC7i9mY8C16fM+vOmaOzt2Hx8pn0MZ6D7r+NiWgYVU8mtWnvnajmSrvOjjR5IGqDIMUfbWkYtO41Aat6zmXoilD4tKv7KJkIoQp5Uu55NIW6+QShqbLtmlUGaPvbL65+aaXqO157O3t3eZhqXtSbIbx4eFhsJGawtjoQkyMlS8JaOzWIZqA0iFzHrkADauiSZl2w5UVfAhjR6R9E+OplDaNC4mx0t5DG6VoypC49Cu7KJkIYRoiGDJliYm1srl4bQJVCYdm+WofgdauDTpDz1k/1tHRkdOIrtclZ+e2pSew1dk3ZUeP1XjKVA0rWzuJSQybW7OHnAJs0/u9vb1ez2nTuGbKl7bfac9Bj9XtuPQruyiZCGEakzu3evHHCIstv5Gvs9TP20aXVYtNMfUdqxLgEIFoXkfdCFmv17dc+vX/d3nBXJ275CXqMTFWKY49FC5hGlOZqmGVcqAyJs3uWvrIZag5Z0xOqDYtaIux8ulDKZoyJC79yi5KJkKYco9+Qko1HdclkDB2/jxF7IPreJrfVx1Zc3zb92xu7q6CULoYhObOKiH1RAguYRpTmaphldILMSbNHnsJ1TCbwRUzGMytKUPi0q9ehAbA1wH4EQA/AeB5vu/nyrxuKylHVyJy215sMaUtgZumjpqRZb0zuI7VlqHXJ5Y+b10VcOrzarVdh/aYrs49R/d1SbiEKWfpW7/GQsqBR6rNvll0hRrWPy79ChGb+wB8HMD7Gp/fDeADAO4H8KrG3/4wgNf5jh2bILSPUVAfQfEpDLVQo6bZuTSjCZ9x0/ytJs3CdrttTUPRJtAhhk7XaQpXCg3fMer38ujo6FZdlsulOTk56TRqa26IfHp6ets5q88B3JoKrX9m82w1p0m7bqOTApcwpS6l6ddYSOWF6NtjFROYPocFUM171Hx+db0RkdtyOtY1tq4XperJ0Lj0K0SYngPgmXVhArAE8CsAng5gH8B7AVys/f3vAnim79gxHquQKaaQ4lpF0ZYorWvxBWdXpWkkhcRYaUeevnxdzd+6NhOujme7n5XB0Dy/doTcxdvk83j6YrJCvKUhI3zbqP6uu+4KalPNZ297Bq59woYAwxpWxehXDGOfZukzxmq1WkWl30m5Ym4sZbFY3Go7oV7Eak/ZUvVkaFz6FSpOxw1hejaAt9b+/epdEQDfA+CrNccNEaa+pwFdnbCvDZa3W/++h65cTbb6+vbca8veG5KZ3WdYuc5tM160L5Au0xS+0XOb0af9bci1NknZvqpz+uqbc8oAA08FogD9imGIWMDmQMm1cCbGwAvtN81VdJVhttls7ljMMhcDqWmcxhqrVZ+P0Rvfb+Y0BenSr67C9A0AXlv79zcB+AEAfwXAewD8MIBXWI51BcANADfOnz+vvpiQDlqfKulaNB6imA5ebVmjCQIPvX5NacuVpTHcqqlB29+McY9S64ai1pBqfi9E5EO8nC5xiBEz7fRkyueqeQYhdesD5DesBtevGPqOBdxu2/PQNT0QfaxEtaUdaZv6buvjfU8xllKam7p3ea9Vfb6Pek45b1UTDG1YhRzTBI74cizb9b2Qq9FdTN02m41XHDS5mrpeX526EeLKtmyrd+UJ83msQqYobfsNakbNoV5OlzjECrmmnvRYlWFYhR63b49V3/nZtJ7lrgZeqLdLow+xephy544cJcViKHqsuuHSr67C1OpKDzmmCRSmoUcomjxUvpeYy3Om6eB1t3wf1+8S6DZxq/aNqn5ru2+2+fgqhYRWqEMMzxRtxiUOXaaiq3raXjCpVk71GWOVMtYH+Q2rwfUrhr49VhrPsut7fXkpXHrq+47PALh48WKyqbWxltCA/9JirHLHHbr0q6sw7QH4MICn4bHgzz8ScLx7AFy7cOGC+mK0AdYpSpUqwXdezV51moBzV6kfJ3WMmU+g6w14vV6rV9NUjd0Wu6EV6iGn7zRTG5pVgbbjt23ZUz9nc5VOW8bmahHF0KsCU8f6IL9hNbh+xdA2VWfLbxeDq73W4zB9HuoutL0kXf3WtTAnpohMZ4PmkGB+7YKs0lYFDhF36MOlXyGi9HoAHwPwBwAeAvCy3efPB/BBnK2uuVd7vHoJDV4fspHWX9i+bVtsMQGpOn9Vl1gjbX9//w6BDm2MId4f30hWMxLfbrcql3XM9F39uaUe9YSOfm2GYe5RWZ3UnhOXMKUupehXDG1ex5TeAVuMVVMjXDrW5aVme0m6NC5GIzR90HXO1IuX+vCQNY1c7aC0JJ3R0LcXV4NLvwYbLbpKyVOB9Re2KwDTZT1r6+zruE3jwXe8tpxKXYPAU+0t5bqfMaPRtiSqFW37K4bEaMUQ2k6b7cxmpOck9VSQS5jGVMY+FWiMf7BWH9S54i6bx9Toiu362vLf2dpcCiNFs5Co63n6nm5srmgO1aGx5KMaelq6DZd+ZRclEyhMQ8+Da8TCJ3yhU1GaPe80ozSN8Ia6VG3X2rzGkNVCNvENEQXbCF5jnPXhQg71rNZfXH14BVIwZo9Vn2Xsweuh59J8J0RXYhf91LG1zWov0dA+2NcAXluXehxrSGlbydn1GCVCj1ViYUrZ4H2Gia+B+TpgJTQxo4YUnhyN8IY2UJfXLrUrOcV0mvbe99EhbcLoMkJ99R1SOJqMPcaqrzIFj1XIuVJ9x/ddV2m2OdvCj8PDQ3NycuLVkra0M6F1SlU0caghGhFjoOXUGQ2TibHqoyAyeD10KwJf47QZWK4AUY1xo/FC2OqbwpOj6Rwxo+Gh5uO7TKf5ri/kemOJMUJLzjtlzLhXBaYuMfoVw5AvEc25NN8J0ZWYGKvmtWu0or4opFmn+hRaHwuENPrT1p9ijM7mPY65ntw6oyF3XJhLv7KLkzFxW9qkzkNiE4Kme7h6iKFpEpor67SBmW1ojQXNfHkJLlUboYLQFmeV02NVXUNI5x/KY5VblIwZv2FVlaltaaM5l+87MZ7w6jf1vTC1/SBkAOXTeddMRrV6MOX2Zq4+3bZwIeZ4ISEsMTpTgp4MjUu/souS6SBMQ4wsNAGNrmJLeOlq1K4tVYwJG8X4RrZDTu3F0OywzQzE9dI2dTtkjFUKcXHVN1XAfQludGNoWE2ZmDYWmu6gToxnJ1bnDw4OzMWLF1v/5gq4txXfAHi73d62ObKrtMVY2TTJNfsTEsReip4MjUu/souS6ShMfQYaVi+xFMeve6l8oyvfaCE2MLp5z+rGSvPffS7vToFNwDSjtT6MxpTi0jZ6j10c0EYpXkqXMI2p0LBqJ2SgEZocd7FY3HGuLoNs7SyERudjfufTV1ufretC0xjSTuva6tx179USZj36xKVfWQUJCWMUunYI18srxbFD6+EjpAOHrtbxba6cG9czyRUbkFpc6i8l7fJ2LSUsVTZm/IZVSv2aOzE5orbb21PFrNdrlceoTedT7XpQ1SPmty59df3OhlaTXNOfGo0pRU+GxqVf2cXJmDQjvq4dwjX9lToxnK9oGnPICK1+PE1ni+nEQ5IyNiAVKcUl5arPNkoZYbqEaUyFHqvuxBoibf3krrvuCtb5VLMe1fFi0wLZsL2Dlsul9TexaTFCNaYUPRkal35lFyUTKUyppwBdc8qu36WO8aq2I9C4zzXetObKRk1ni+n4XeOLfL+v/91Vv1zTlSnFRduuu3jDSoiJcAnTmMoUDatUwci+GJ8+wzhCdD5VfsTq+LG/b0vc7PuNDY0m+Y4fO8hnjNUIhamvoHXbJrmarWxCXNFtnan+35AG6usYzS0O+vBYde1Yvt9rn3fOqcqU4qIR+SqBYF+G7BC4hGlMZWqGVaq27DrOEAuPQnTe5hGKWf13cHAQvMFxW//WpBVyGT6a5+jTGm0Qewl6MjQu/couSiZCmPoc5bS5l1er1R3B3JrcLppSLd91/Ta087h+rzFiXPcm5HloPSq+32uedwkjpFTiYrve5XJ5y4DvuudjCbiEaUxlaoZVKu+r6zhdNdyX5ypG59vO0bZdmaakTMfQRfN8mjQWbS0Rl35lFyUTIUxDb2sDPBaUmDKRW1U0MVwufK7ntoRxts7mug7bMl7teUOfZ/V73zlcRkzbtZY+uvIZvymnHXPiEqYxlakZVqniBV3H6aLh9X7cl9erPhvR1zula6kMyy77iWrv4di0ZQhc+pVVkBC5qiakse/t7SVryC7LvU9jr8qx0uWe2OIbmsaF6zrqv03ZGWM9Vr7jt9VT430sAZfxN5VVOC5hGkOJ1a/SKcVj1Rxw2nICpk4WXT9PjkG89p3gq7uWlIPkOeHSr+ziZEx/MVZVfFEzDsqWT0rjXraJi08oXMfWeKw0RoQvQ68rvqH6m0ZUU7uPY2KsNMcPMcBzj8i0XrSuS6NLwiVMYypT81iVEmO1Xq/VnuVm/6lW/nXV+ZweK1uMVdc8iLZ75jJSx6YtQ+DSr+yiZCKFSeOm9VnZtmkiTcxSs7O3/a6ZKdsmNJp5fM2IQTN6c40Wq3p2CXgcYlWg9vgho82cIzLti8zVNkv0uvlwCdOYytQMK2PKWBWYqk/G6nw1EA/dUqZeusRa1e9PfbovhZa1Xf/+/n5rfUtLDl0KkzSs6qRyXVf4YpZsL7TT09NbHWC5XLZuTWMTGp/IhF5LTHxD1SFjAx5LG9WEjDZDRsdD1bN5P23fa678HAs0rIgx+fREq/OLxeKWwVHpumZgX+lJjFHlunbfeTWro0O1kdzJ5A0rlyco9mUZ2gmGcJ+H4BKrrkKWqo59o42xalvaPOT1aGOmphJbVUHDihgTpyepPGoxRo9mRbgvtKJutMRcu8vb1sxd2MZYvPklM3nDypj2OfYuL/8Qi746ZxdjxXUtMaIRGt8Qakj4PG+lrLjTrArMHVfQ1WNVmqdQCw0rUhGiGykHdiE67+pzLj30Td/bpvw0hqW2fl2ue6z60jfFGlbocVWNreFop3y0AfLVMUvzJjTjrdo26AwRsth7VqInq4lr9NZHbEnbd2NjrGLvbwkG8NgNqz71i9jRDDC6aJZ2MK2lHiJSr2t9RiU2N512FXfb4LItxirWc1+CngxNsYZVVfoY8Wldna6GU28srpV7R0dHRXkTUr+Atccq6R6E4Bq9DTWdG/Ii6NObOSRjN6yqQo/VsGhiREPat2bKLlbTuixY0pzHVu8qRY9v5sLnzdcaVSXoydDM0rAK6Sia4DxfoGNJjSulgRNyrFivXR/esxB8o9bQ+1a6gVlK/WhYkRh87dc1W+FC+75IEVKiSbFT188qTVDbVKNNdzcJ4mm7XGMpetcXszSsQl28mo7i+n11zhQjgOZ1hP4+5bRkyLFiOliIQdqn8eoynFNmnC6BUupHw4rE0KYDInJrFXbs1L5vMO4ycKp6NbW6rwSjdd1z1XeIvl6KngzNLA0rY87mtrUNW2Nd24KcmyOhumu5LTmd1hAINTp805Z9e6xiDJ+Q4/c1MuoaCDpUPVNRSv1oWJFY2rRdsxLPN8izvS/qiaXbzmmLWbLlsQrxWPmuxbX4hh6r/pitYRUyHVgZQS6v0Ha7vSO4r7m0VeMp0zY4bYPVnHOIGKvq+yEetpDRTh8jI8290+7w7jpmdc80uc76ppRpaxpWJBaXNvrCNlz61GawtaVo0RowNi1ui7Ha3993LoRq0722d1JVqnxbNn1LpT+l6MnQTNaw8r3EY92wXQwHTQfTGgJaQ8I1j58iFqmvuKbQkWUfIyOtIKZIR3F6emoVQNfv+iB3XJsxNKzGTs6VYD5ttHlxNDMIzevy7WbhmnJr0+R6fJTt/mn2P/QZdG3pHNrK0dFRrxn2cx6rTyZpWGmsZFtj0nSCkJd1vSFoOldqj9XY5rh9XqKuMVYhHTPE+O7q2ra5/5fLZdD1DUmfdaJhNV5yt1WfNtrisEL7tc/7pTFwQjV5u/Xv+xrzrtFMP05Zb1JTrGGFDnlgNEaHbR6+vkGnr1H6HmhokHy9kfgMAG0j68OT0yc+T1UX70loxwxx4Xc1VF3HdtUl53Pss04uYRpD6aJfYyV1PGIsrhirZl2roPNQrQ8xcNryUMXeH9v9reK1YrOma38zVb1JjUu/souTMXEjPt9LyrdyRGsQxQZf+zpvSu/LmKx8Y/r1sGk7pmuBQczIVoPPY1Wi57HPOrmEaUxlLh4rjWYO0VZ92h5T7zbNDNF24PYYqdiEn8a4+1xonUI9VlPWm9S49Cu7KJlIYbKttlgsFsaY+FwnoS9U7TGapLbMxzIvbUy/oxJNx3RNE2w23bdDsuGLsSpxtEaPFQ2rCq2XP1c9bOeO1foQA6bt97Ga7Lq+EG9VU7Ns+jMXvUmNS7+yi5KJFCafEeN7wca6U+u4luf6GsWYLPPU9Olh03RMzXf6MlRdqwJL9DwyxoqGVYVP64Zqq6HaGaP1Wm3vQ79dfU5rJNo0q64/ImL29vZmozepmaVhNYTHSnOMqW3/koo+DZfmPW2mxNAKc191dB1XE0M2tGeSqwLHaVh1fW6hq+NKCVzXfj/2/bBarXrfuN327LbbrWpKz/a8m8et71fY9u8cBs1YZl8mZ1i5XJpVss4hYqxcIxpNTqyxWObGjKOxb7ftOV329vZUsRNNj1Ufz6fLcUN+6xPQEp4fDav+6Np+237vCs4eckDom8rXLGjxab0ve3sKfYjRVO10Xlt9fHXW/L10TRmSyRlWLou9OadsWzmy3W5VuUJiAte1IlOid8JWzzEYga5RZqjR1JdH0VfHFO1N8yIp4fnRsOqPru03xMuToy25Fp+0GRShWq8dfMUaGrGaGpqxvV5f2zVVubVc1zwWTRmSyRlWrodbx9ZQ1uu1agTjE6E+DY6SjJm+jIzU+DyIdXxGa18xcF3iVLomjC3t+dGw6o+u7dfVTqupsNwDPmPi8llp+kKo/oZ+P1ZTQ4yq5vN2PVPXPQpZjZhbU4bEpV/ZRcn0aFjFZl4PMWL68iqVZMyMJdBe67Hqcqw+PVa+c2jrFNLuc74UXcI0plKiYdW3x6qUF6hPmzT9TTudnnL2IlZT+/BYVcW1x2yIpswFl35lFyWT2LCqdwSN+7etYecehRlThjHjW4VSirhW2GKs9vf3g5+nJgGhpj5NYdaMoF2rULskjA15qQwBDav+6CPGaggdCh2o+qa3usTBhhCq17HB77YYq5OTE+s+tq5p0zY9aGszWk0RkezvzqGYnGFle8jNRuPbPLNkQ6EPj0mIaPmENVSk+/DqtZ2nKVihGyhXxwlJQKg9hnbvLtcz1txL7fRH7nZPw6pfuva77da+Aq2PNhNjDLra+sHBQe+r9yo0U5LVs1iv1633dbVaeeO0ttutOTw8vPWbxWJxa7Pl5rtuf3+/NSefSwdsbSZEU0p6j/ZJsYYVIreEcK0KaRZbI/YZCtoXmK8T1P8+RHBjquP5XvxdRr4pvSQhQaxaUhi1mmPY7k2K1TZtbc/2PHNN6Y7dsIrVrzExZKxnbL9zGYBt8bSabcVs3uaqjtX5XJ7o+kBKY5QcHR056+ryors8d7Y+33avfO+k+j0pTU+GpljDqiqxmzBrH3KoO1gjJr7vpFhFkdLTEypaqaYi+/C8VWjuccx5Ulx7bK6svrK+G1NW3J4x4zesqlKqxyoVQ3mcu/Q7129DvDDVwKb5+f7+vjXdRNMT3TyP6/2kKdWxbNdYnTP0uG3evBCtKU1PhsalX9lFySQSJtdDDm0Amu/7vqPtTEM1wlDhSdVpbNedYlSjuccx5xnKYxXyuyo/WxeG9D5ocAnTmMrUDauh6NLvYn4b6uWJ1fEui6gqDfM5D0KvJea92KQ0PRkal35lFyWTSJhCXbFdl7X7vqPtTNoXf9dRo+uFrR21xQRvu0ZZXdHcY9t5XPczhWDEHsOXmLArQ3kfNLiEaUyFhlUauvS7mN92NXjadDxkkNqsaz12qqnRrrrWt6gJ0XFXfUKeWSl6MjSzMKyM8b8stQ2gNI9Vny96V3BnX8ZcqpUjvnvsWkatmertKhgxx3Bd01DezaGgYUWadOl3ob9N7bEKmVpsnm+7tScxXa/X6vdJZVzVr992X2zXuVwu4x/gjJiNYZWKUmKsKlJNy7V1sD7TOrhGWSlwLWJwCWuq+9kHrpHk1IJCaViRnKSOsXLpiqZf+8I1Uq/Kcx2D+KFhFYFm9OP7TltgcsxorE/jp08jI8WxQ++x5p72bfB1Zagl4rmhYUVyk2pVoDHxyUp9sxx1j1X9vF0HYCUPMMcADauR02cH6DMAseux+6pb31OUXZlLUCgNKzIlfDodM8vRloux+o1r6lDDXHSmL2hYjZy+O0CfAYhdjt2XQdl3UH0K5hAUSsOKTIkUsZvNv7u8110NK019iJ3ZGFZDNZIcjXGOHaDPKdCubvQ25viMukDDijQpoQ8NGUDvw6WBGn0s4X5OlVkYVkO5Nek+HY7S47/qsF2EQ8OK1CmhD5VQhzouneo69Ui6MUnDKsRlmpK+451KnJLLRcnxX01KDgQt9dnTsCqTXO2lhD4UU4cUC51cx47NzzjWd9VYmJxhFbL0NPUS9b6mp8ZkRAzJWIzNPqctu1Dys6dhVR4520sJfSi0DilS8/hw6ZTrb2N8V42JyRlWNkt8zB4r23FTbGVSwkhwSrSJWR/3OIXhV/Kzp2FVHjnbSwltNbQOmu8PfV0uPWo7b6jOlPCcSqBYwwqRu8NrtyIYU4yV6zr6DIAkYdief+rNk1O1s5Kf/dgNq1j9Kpmc7aUET0hoHVy6XTHkPfXN5qTwppWsKUNSrGFVlVQeqyqRWuwIv27ptyWAa34v5fSUaxuFrvP7HGGkw3UvU7Yf23kWi8Vt7d3W9kJHrTkYu2FVFXqs0jH0yu62vhpSB5dhVR3DtRFyanyzOUdHR7f2Htw4YpOr+3J6etp6z0rVlCGZnGHVx8jGZekPMWpydYYu8/sh3yN+fKM1373WPgutV3a1Wqmed4nPnoZVecxBK1JqvaaPDtkHU24sXZXT09PRaMqQTM6wMubOkU3sdjEVPku/b2s8ZiRg+81yuVRt0VACpdbLhm9E3/XvvvN0/W1J95iGVZkM2Sdz9P+UWt9lw2YXsfclRDe0ZblcjkZThmSShlWdFKMsn6Xf9/zxdrtt3eRzf3/feh2a0UnJI4kxjo59dfZ5tLTxCV1Wvo4lBoKG1bzJ1f9Tav3p6WmQkaK5vi73JUQ3QspYNGVIJm9YpYgL8Fn6y+Wy9w6/3d6+/5MrhkZTZ+19yOU1yh3PEYvrfqXyWDXPExKDN5b7SsNq3uRqpz7dDF2JfXp6eqt/LpdLc3h46H2X9BkTq9WN+vW67onLY1WapgzJ5A2rFNa0xtIvzZuiHZ247sPc89akJlWMVdtxm5uxAvoYq9LarjE0rOZOrv7v003XLEGK4/v6Zcr7EhIbZfO+2WKsStSUIZm8YWVb2RAaa+Rb9VCiha4ZnbjqnHMkMsS5c3jjfOeMrVOIR3MMsWs0rObLdruNXi2nadvaPtiXzruuz3cujS6G9O/6d9fr9W2rApu/a3rf2lYFlqwpQzJpw8o2km+zzENyDY3RmxIzqphy3hqOssqGhtU86bIqT9OnQ/p9n/oXO6PQl9ebpGXShpU2zqiywLWjk7HOKYeOKnJfZ5+joNzXRtzQsJontn6piWPV9OmQft+3RsTOKHSJ4yTDMGnDKlXejmYjnsuooO06V6uV011sO07V4V3JMYckZDRKN/fw0LCaJ128RF30XkRa0/QMpfM2D9Z6vTanp6fqaX7XPaB+DcekDasQj5WrMbZ1rLm8bJtz8M2p1Rj3fAnGqHZkNxcjujRoWM2TLh6XLnq/Xq+t21ENpfPNOElXsQXRa+4B9at/Jm1Y2TwumrirNqOq3glLYUgDL0b0fB09l4taazDRtZ4HGlbzJEeepoODA6tBk1LrNVodYhzapgk194D61S+TNqyMaW/M2+1W1fBcfy/B4h/amxLjptcYr7nQCN0YFypMARpW86XLYLH+W432VMd3fT+Fnmq1OmQ607Wdme8eUL/6ZfKGlQ2X4VRZ85rv5GRob0rM+XzLiku4jy7oscoDDSvSlRD97lvrtTrS1WMVe16SFpd+LTBhrl69iv39/Ts+X61WuHr16q3v2HjwwQd7q5sWWx36qtvVq1dxcHBw22cHBwfO+/SZz3zG+jffb0sg5pqnxPXr13F8fIzFYoHj42Ncv349d5UIUaHR+Pp3baTQU61Wt+mNDY0Glahfs9cUm8U1ZOlzxKdJqmibey/B4s8xGkmVsmGIbYBSMZeFCk1yBu6DHiuSgJDEuX1qfYhWa9IwhMR+laRfc1kM5NKv7KJkChCmkhtCyXWrGEMdSTs5pxFoWJGh6VOrYo893Lg6mQAAEtpJREFUNf2cy9SkS79GOxV4/fp1nDt3DiICEcG5c+ei3Y2XL1/GtWvXsNlsICLYbDa4du0aLl++rK5Ln27Pxz/+8bf+f71eW+s2tPv1la98Jfb29vCSl7wEn/zkJ3F4eBh0/2bvLi6AoaeaSV669rmx99lQrQ+93qZWv/SlL8W9997r/P073/lOfPKTn7z178PDQ9XvSoWagnF6rLbbrVmtVndYxF03z4yhlBHQ0KMe14adGqY2Shsr9FjNx2PVtc/Nrc921d+2tD/N39t0dLFYjPY+02M10qnAklby9dmIQo49dGO2xQUsl0vV7+fS+UqHMVbzMay69rm59dkU+uv7vXaj5jHd57kY4IMaVgCeDuB1AN6g/U2oMJWUu6PPHEghxx46F5NLADQwd1Q55Ap8LdWwCtWwsRhWXfvc3PpsCv31/V77m7Hd55KC6fvCpV+qGCsRuU9EPi4i72t8freIfEBE7heRV+HsrfphY8zLNMeN5fz589a/fc7nfE6fp74DW10Wi0XneXHbsds+D/luCpbLZdDnTYauL7Fz+fJlPPDAA3j00UfxwAMPqGMLx0RpGpaDrn0u9vdjjcuyXVfbOyZEt+rf1epl6DlyMwdNcWKzuMztI7jnAHgmgPfVPlsC+BWcje72AbwXwMXa33vzWNlirIDh46z63CePMVZkymBAj1WfGjYWj1WOGKsx9/OQWF7GWM0Pl36FCNNxQ5SeDeCttX+/GsCrTaAomUhh2m63dzS+qgw9F73dbq1z5V3rEuJSHdr9enp6euu6l8ul2qiqmIO7mNgZ0rAyPWrYWAwrY7r3uVQ57sYSLxSS98q2tZrvfrXpKLWxfFz6JWd/9yMixwDeYoz54t2/vwHA3caYv7j79zcB+G8BfBeAqwD+FIDXGmP+luV4VwBcAYDz588/6+bNm6p61FksFmirv4jg0UcfDT5eF0qqCyFjQETeY4y5NOD5jpFIw1Lo1xwYuy6Ovf6kP1z6lTyPlTHmt4wxrzDGfIHNqNp975ox5pIx5tITn/jEqHOVFKdTUl0IIfFoNCyFfs2Bsevi2OtP8tDFsPoogKfW/v2U3WeDUdIeSSXVhRCiIruGTZ2x6+LY60/y0MWw+gUAXygiTxORfQAvBvCmNNXS0TVj+lTrQghRkV3Dps7YdXHs9Sd5UMVYicjrATwXwDkAvwHgu4wxrxOR5wP4PpytrrnPGBNkxovIPQDuuXDhwss/9KEPhdadEDJihoyx6kPDqF+EzBeXfqmD1/vk0qVL5saNG7mrQQgZkKGD1/uC+kXI/Bg0eJ0QQgghZK7QsCKEEEIISURWw0pE7hGRaw8//HDOahBCSDDUL0JIG1kNK2PMm40xV57whCfkrAYhhARD/SKEtMGpQEIIIYSQRNCwIoQQQghJBA0rQgghhJBEMHidEEIioH4RQtpg8DohhERA/SKEtMGpQEIIIYSQRNCwIoQQQghJBA0rQgghhJBEMHidEEIioH4RQtpg8DohhERA/SKEtMGpQEIIIYSQRNCwIoQQQghJBA2rmXP9+nUcHx9jsVjg+PgY169fz10lQgiZJNTbebCXuwIkH9evX8eVK1fwyCOPAABu3ryJK1euAAAuX76cs2qEEDIpqLfzgR6rGXPvvffe6uQVjzzyCO69995MNSKEkGlCvZ0PTLcwYx588MGgzwkhj0H9IiFQb+cD0y3MmPPnzwd9Tgh5DOoXCYF6Ox84FThjrl69ioODg9s+Ozg4wNWrVzPViBBCpgn1dj7QsJoxly9fxrVr17DZbCAi2Gw2uHbtGgMpCSEkMdTb+SDGmNx1wKVLl8yNGzdyV4MQMiAi8h5jzKXc9egK9YuQ+eHSr9F6rFz5QJgrhBBC4ilJQ0uqCyEaRpnHypUPBABzhRBCSCQl5VsqqS6EaBnlVODx8TFu3rx5x+ebzQYArH974IEHoutICEkLpwLLxKWvQ2toSXUhpI5Lv7J6rETkHgD3XLhwIeh3MflAmCuEEJKSWP0qnZLyLZVUF0K0jDKPlSsfCHOFEEKGYKp5rErS0JLqQoiWUQavu/KBMFcIIYTEU5KGllQXQrSM0rBy5QNhrhBCCImnJA0tqS6EaBll8DohZPwweJ0QMlYmmceKEEIIIaQ0aFgRQgghhCSChhUhLcRmew75XY6M0kNcFyFkeNr6qPazlOdMyWh1xxiTvTzrWc8yhJTCdrs1BwcHBsCtcnBwYLbbbbLfxZ6jC0NcVwgAbpgC9KdroX6R3LT10dVqZfb392/7bH9/36xWqyR9uW8Ny6GRIbj0i8HrhDSIzfYc8rscGaWHuK4QGLxOSBpsfVRLTF/uW8NKz7rv0q+shlUtc/HLP/ShD2WrByF1FosF2vqFiODRRx9N8rvYc3RhiOsKYeyGFfWLlIKtj2qJ6ct9a1gOjQyh2FWBZqKZi8m4ic32HPK7HBmlh7iuOUH9IqXQtS/G/L5vXRiz7jB4nZAGsdmeQ36XI6P0ENdFCBmetj66Wq2wv79/22f7+/tYrVa3fRbbl/vWhVHrji34asjC4E9SGtvt1mw2GyMiZrPZqAMmQ34Xe44uDHFdWsDgdUKS0dZHtZ+lPGdKcmikFpd+MXidEJKFscdYVVC/CJkfxcZYEUIIIYRMCRpWhBBCCCGJoGFFCCGEEJIIGlaEEEIIIYmgYUUIIYQQkggaVoQQQgghiaBhRQZntDuWJ4b3gRDSpG9doO4MgC3B1ZCFCfbmQ+k7lg8F7wMThBLSpG9doO6kw6VfTBBKBqX0HcuHgveBCUIJadK3LlB30lFsglARuUdErj388MM5q0EG5MEHHwz6fKrwPowf6hdJTd+6QN0ZhqyGleHu8LNjzDuWp4T3YfxQv0hq+tYF6s4wMHidDMqodyxPCO8DIaRJ37pA3RkGGlZkUC5fvoxr165hs9lARLDZbHDt2jVcvnw5d9UGhfeBENKkb12g7gwDg9cJIVlg8DohZKwUG7xOCCGEEDIlaFgRQgghhCSChhUhhBBCSCJoWBFCCCGEJIKGFSGEEEJIImhYEUIIIYQkgoYVIYQQQkgiaFgRQgghhCSChhUhhBBCSCJoWBFCCCGEJIKGFSGEEEJIImhYEUIIIYQkgoYVIYQQQkgiaFgRQgghhCSChhUhhBBCSCJoWBFCCCGEJGIv9QFF5BDADwH4FIC3G2Oupz4HIYT0BTWMENIFlcdKRO4TkY+LyPsan98tIh8QkftF5FW7j/97AG8wxrwcwAsS19fK9evXcXx8jMVigePjY1y/Ti0khJwxBg2bK9RuMjW0U4E/CuDu+gcisgTwgwC+BsBFAN8oIhcBPAXAR3Zf+0yaarq5fv06rly5gps3b8IYg5s3b+LKlSvsoISQih9FwRo2V6jdZIqoDCtjzDsA/Hbj4y8HcL8x5sPGmE8B+HEALwTwEM6ESX38rtx777145JFHbvvskUcewb333jvE6QkhhVO6hs0VajeZIl1E48l4bFQHnInRkwG8EcDXi8j/CeDNth+LyBURuSEiNz7xiU90qAbw4IMPBn1OCCHooGEp9WvOULvJFEkevG6M+S8AvkXxvWsArgHApUuXTJdznj9/Hjdv3mz9nBBCQtBoWEr9mjPUbjJFunisPgrgqbV/P2X32eBcvXoVBwcHt312cHCAq1ev5qgOIWQcFKNhc4XaTaZIF8PqFwB8oYg8TUT2AbwYwJvSVCuMy5cv49q1a9hsNhARbDYbXLt2DZcvX85RHULIOChGw+YKtZtMETHG78UWkdcDeC6AcwB+A8B3GWNeJyLPB/B9AJYA7jPGBA0zROQeAPdcuHDh5R/60IdC604IGTEi8h5jzKWBzpVcw6hfhMwXl36pDKu+uXTpkrlx40buahBCBmRIw6pPqF+EzA+XfnEpMSGEEEJIImhYEUIIIYQkIqthJSL3iMi1hx9+OGc1CCEkGOoXIaSNrIaVMebNxpgrT3jCE3JWgxBCgqF+EULa4FQgIYQQQkgiaFgRQgghhCSChhUhhBBCSCKS7xUYQpVgD8DvikhMhr1zAH4zba16Z4x1BsZZ7zHWGRhnvWPqvOmjIkOh1K/SniXr44b1ccP6PIZVv4pIEBqLiNwYW4LBMdYZGGe9x1hnYJz1HmOdh6C0+8L6uGF93LA+OjgVSAghhBCSCBpWhBBCCCGJGLthdS13BSIYY52BcdZ7jHUGxlnvMdZ5CEq7L6yPG9bHDeujYNQxVoQQQgghJTF2jxUhhBBCSDGMwrASkbtF5AMicr+IvKrl73eJyE/s/v5uETkevpZ31MlX578uIv9RRH5JRP6ViBSx9NxX79r3vl5EjIhkX5GhqbOIvGh3v98vIv9k6Dq21MfXPs6LyM+KyL/btZHn56hno073icjHReR9lr+LiHz/7pp+SUSeOXQdS0REvlRE3iUi/15EbojIlxdQp28Vkf+06w9/O3d9AEBEvn2nKecy1+N7d/fml0Tk/xKRz85QB5UOD1SXp+60qNLPv5qzPhUistzp41ty1+UOjDFFFwBLAL8C4OkA9gG8F8DFxndeCeCHd///YgA/MYI6/wkAB7v/P81dZ229d9/7LADvAPAuAJdKrzOALwTw7wD84d2/P3cEdb4G4HT3/xcBPFBA+3gOgGcCeJ/l788H8M8BCICvAPDu3HUuoQB4G4Cvqd2jt2euz58A8C8B3LX7d9b+sKvDUwG8FcBNAOcy1+V5APZ2//89AL5n4POrdHjA+nw+gGfu/v+zAHwwZ31q9frrAP4JgLfkrkuzjMFj9eUA7jfGfNgY8ykAPw7ghY3vvBDAj+3+/w0ATkREBqxjE2+djTE/a4x5ZPfPdwF4ysB1bENzrwHgb+JMcD45ZOUsaOr8cgA/aIz5/wDAGPPxgevYRFNnA+AP7f7/CQB+bcD6tWKMeQeA33Z85YUA/pE5410APltEPn+Y2hVNac/yFMB3G2N+HyiiPwDA3wfwHTi7V1kxxrzNGPPp3T9zaLNWhwfBGPMxY8wv7v7/9wD8MoAn56oPAIjIUwD8GQCvzVkPG2MwrJ4M4CO1fz+EOx/qre/sOsTDANaD1K4dTZ3rvAxnI/3ceOu9m955qjHmp4esmAPNvX4GgGeIyDt3UzJ3D1a7djR1fg2Al4jIQwB+BsC3DlO1ToS2+7nwbQC+V0Q+AuDvAHh15vo8A8BX7cImfk5EvixnZUTkhQA+aox5b856WPgLGF6bi+1HuzCbPwbg3Xlrgu/DmSH+aOZ6tJJ1SxsCiMhLAFwC8Mdz18WHiCwA/D0A35y5KqHs4Ww68Lk4G32+Q0S+xBjzO1lr5eYbAfyoMebvisizAfxjEfliY0yRQjJ3RORfAvi8lj/dC+AEwF8zxvyUiLwIwOsAfHXG+uwB+BycTdd+GYCfFJGnm938Sob6fCfOpt8Gw1UfY8z/vfvOvQA+DeD6kHUrFRE5AvBTAL7NGPO7GevxtQA+box5j4g8N1c9XIzBsPoozubfK56y+6ztOw+JyB7O3O2/NUz1WtHUGSLy1TgTlj9eueUz46v3ZwH4YgBv3820fh6AN4nIC4wxNwar5e1o7vVDOIv3+QMAvyoiH8SZofULw1TxDjR1fhmAuwHAGPNvReRxONsXq4RpGxuqdj9FjDFWQ0lE/hGAKuD3n2KA6QtPfU4BvHFnSP28iDyKs7b1iaHrIyJfAuBpAN6705SnAPhFEflyY8yvD12fWr2+GcDXAjjp0+C0UFw/EpEVzoyq68aYN+asC4CvBPCC3YKexwH4QyKyNca8JHO9bjGGqcBfAPCFIvI0EdnHWXD6mxrfeROAl+7+/xsA/OsMnaGOt84i8scA/AMALygkxgHw1NsY87Ax5pwx5tgYc4yz+IOcRhWgax//DGfeKuxWHD0DwIeHrGQDTZ0fxJmnAyLyRTgTkN5efIl4E4A/v1sd+BUAHjbGfCx3pQrg1/CYR/pPAojZcD4l/wxnAewQkWfgLEA6y0a2xpj/YIz53JqmPISzQOnejCofu1CB78CZtj3i+34PaPRhMHbxyq8D8MvGmL+Xqx4VxphXG2OesmsvL8bZ+74YowoYgcfKGPNpEfnLOFsxsgRwnzHm/SLyNwDcMMa8CWcP/R+LyP04C659cb4aq+v8vQCOAPzT3UjtQWPMC7JVGup6F4Wyzm8F8DwR+Y8APgPgfzLGZPNoKuv87QB+RET+Gs4Cer8582ABIvJ6nBmo53axX98FYAUAxpgfxlks2PMB3A/gEQDfkqemxfFyAP/7zpv+SQBXMtfnPgD3yVnajE8BeGnutlUYPwDgLgD/YqfN7zLGvGKok9v0Yajzt/CVAL4JwH8QkX+/++w7jTE/k7FORcPM64QQQgghiRjDVCAhhBBCyCigYUUIIYQQkggaVoQQQgghiaBhRQghhBCSCBpWhBBCCCGJoGFFCCGEEJIIGlaEEEIIIYmgYUUIIYQQkoj/HzufkLHjUJt7AAAAAElFTkSuQmCC\n",
"text/plain": [
"<Figure size 720x360 with 2 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"pred_likelihood, pred_logit, meas = list(), list(), list()\n",
"\n",
"for domain_tuple, domain_measurements in measurments.items():\n",
" for pseq, (value, units) in domain_measurements.items():\n",
" if units == 'nM' and float(value) < 1e3:\n",
" # Returns both the likelihood and the logit \n",
" \n",
" likelihood, logit = sequence_predictions[domain_tuple][pseq]\n",
" pred_likelihood.append(likelihood)\n",
" pred_logit.append(logit)\n",
" meas.append(float(value))\n",
" \n",
" \n",
"corr_likelihood = stats.pearsonr(pred_likelihood, np.log10(meas))\n",
"corr_logit = stats.pearsonr(pred_logit, np.log10(meas))\n",
"\n",
"## Violates normality so Pearson isn't super accurate... Spearman is equivalent between two.\n",
"print(\"Correlation (Pearson): logit - ⍴ = {0:.3f} (p = {1:.3f}); likelihood - ⍴ = {2:.3f} (p = {3:.3f})\".format(*corr_logit, *corr_likelihood))\n",
"print(\"Correlation (Spearman): ⍴ = {0:.3f} (p = {1:.3f})\".format(*stats.spearmanr(pred_logit, np.log10(meas))))\n",
" \n",
"fig, axes = plt.subplots(1,2, figsize=(10,5))\n",
"axes[0].scatter(pred_likelihood, meas, c='k')\n",
"axes[1].scatter(pred_logit, meas, c='k')\n",
"axes[0].set_yscale('log'); axes[1].set_yscale('log')\n",
"axes[0].set_title(\"Likelihood\"); axes[1].set_title(\"Logit\")\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"With lines of best fit:"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 720x360 with 2 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"ticks = [0,1,2,3]\n",
"tick_labels = [\"$10^{0}$\".format(t) for t in ticks]\n",
"\n",
"fig, axes = plt.subplots(1,2, figsize=(10,5))\n",
"for ax, pred_subset, title in zip(axes, [pred_likelihood, pred_logit], [\"Likelihood\", \"Logit\"]):\n",
" sns.regplot(x=pred_subset, y=np.log10(meas), order=1, \n",
" color='k', line_kws={'color':\"#EA157A\"}, scatter_kws={'alpha':0.1}, ax=ax)\n",
" \n",
" ax.set_title(title)\n",
" ax.set_yticks(ticks, tick_labels)\n",
"\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Models\n",
"\n",
"To use a model fit to these data, see the get_Kd function below."
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"2.7021873404460734"
]
},
"execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"## Models:\n",
"LinearFitModel = namedtuple(\"LinearFitModel\", [\"slope\", \"intercept\", \"rss\", \"mean_error\"])\n",
"def make_model(predicted_values, measured_values=np.log10(meas)):\n",
" slope, intercept = np.polyfit(predicted_values, measured_values, deg=1)\n",
" \n",
" # Model quality estimators\n",
" estimator = slope * np.array(predicted_values) + intercept \n",
" rss = np.mean(np.power((estimator - measured_values), 2))\n",
" mean_error = np.mean(np.abs(estimator - measured_values))\n",
" \n",
" return LinearFitModel(slope=slope, intercept=intercept, rss=rss, mean_error=mean_error)\n",
"\n",
"logit_linfit_key, logit_linfit = \"Logit-Regression\", make_model(pred_logit)\n",
"likelihood_linfit_key, likelihood_linfit = \"Likelihood-Regression\", make_model(pred_likelihood)\n",
"\n",
"regression_models = {\n",
" logit_linfit_key: logit_linfit,\n",
" likelihood_linfit_key: likelihood_linfit\n",
"}\n",
"\n",
"def get_Kd(aligned_domain_seq, aligned_peptidic_seq, regress_key=logit_linfit_key, regression_models=regression_models):\n",
" \"\"\"\n",
" Compute a K_d from two regression models, based on either the likelihood model or the underlying logit.\n",
" \n",
" Inputs:\n",
" - aligned_domain_seq: an aligned domain sequence (i.e. aligned SH2 domain)\n",
" - aligned_peptide_seq: an aligned peptidic sequence (i.e. centered phosphosite)\n",
" - regress_key: the key into the regression_models dict. Can be {\"Logit-Regression\", \"Likelihood-Regression\"}\n",
" - regression_models: dictionary of linear regression models\n",
" \n",
" Returns:\n",
" - a log[K_d] predicted from the input sequences. \n",
" \"\"\" \n",
" likelihood, logit = _compute_interaction(dseq, pseq, model, amino_acid_ordering)\n",
" linfit = regression_models[regress_key]\n",
" val = logit if regress_key == logit_linfit_key else likelihood\n",
" \n",
" return linfit.slope * val + linfit.intercept\n",
"\n",
"# Example:\n",
"\n",
"data_pt = training_data[0]\n",
"domain_tuple = tuple(data_pt[:2])\n",
"dseq = alignment_mapping[domain_tuple]\n",
"get_Kd(dseq, align_phosphosite(data_pt[2], 7, 7))\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This clearly isn't great regressor. In part, the challenge with these data are their reliance on a noisy experimental source. For this estimator, the mean absolute error in predicted $\\log K_d$ is:"
]
},
{
"cell_type": "code",
"execution_count": 512,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Mean error (logit): 0.331 (log[nM])\n",
"Mean error (likelihood): 0.327 (log[nM])\n"
]
}
],
"source": [
"print(\"Mean error (logit): {0:.3f} (log[nM])\".format(logit_linfit.mean_error))\n",
"print(\"Mean error (likelihood): {0:.3f} (log[nM])\".format(likelihood_linfit.mean_error))\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"If we compare this with where most of the data comes from (Koytiger, et al (2013), PMID: 23358503), the allowable mean absolute error ($\\log[nM]$) in the underlying estimate is, by my estimate, ~0.1 [1] with an FPR of ~3.8% (their estimate). Although certainly our predictor remains fairly noisy, a pretty large chunk of this noise is associated with the underlying experimental variability.\n",
"\n",
"[1] The method fits an equation ($F_{obs} = F_{max} * [\\mathrm{peptide}] / (K_d + [\\mathrm{peptide}])$ where $F_{obs} / F_{max}$ are the observed / maximal (saturated) fluorescence at a given concentration and $[\\mathrm{peptide}]$ is the peptidic concentration). This is fit over 8 peptidic concentrations and in quadruplicate. The paper allows for an $R^2$ value of $0.9$ for an experiment to hold. Simulating these conditions leads to that estimate. "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Discussion"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"When utilizing this model, it becomes important to consider the predicted change vs the likelihood of significance of the change. Specifically, if we consider two peptides targeting a domain, $\\log$-fold difference in predicted $\\log[K_d]$ on the order of 0.1 or 0.2 are well within the range of error of the predictor (similarly, the likelihood predicted in the model is not sensitive to small changes in probability). However, large changes in the likelihood (*i.e.* 0.3+) likely correlate with meaningful changes in $K_d$.\n",
"\n",
"When chooosing between the logit and likelihood model, the major consideration is in the non-linearity associated with the likelihood model. Specifically, this function (the \"sigmoid\" function) compresses the tails of the energy model. This is slightly non-physical, however, can be helpful in practice (the likelihood model has slightly better performance). Regardless, both models have approximately equal absolute errors.\n",
"\n",
"One challenge in fitting a specific $K_d$ to the entire model is that a large fraction of the data are concentrated in the *$\\mu$m* range (*i.e.* approximately $10^3$ *nm*). Although physically realistic (*i.e.* relatively few peptides bind at *nm* concentration), this does limit the applicability of the predictor in low-$K_d$ regions. "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Regressor"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Regressors: One option for directly fitting to binding data is to convert the problem from a classification problem into a regression problem. The underlying principles should be the same. \n",
"\n",
"A big problem with fitting a regressor here is the dataset. As described above, most of the \"negative\" values do not have an associated $K_d$ (which makes sense). A putative regressor is then fit on a relatively small dataset ($\\sim 2,601$ data points). Consequently, some of the benefits of the likelihood model (*e.g.* leveraging a large peptidic space, including negatives) are lost."
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {},
"outputs": [],
"source": [
"from scipy import sparse\n",
"from sklearn.linear_model import LassoCV, LassoLarsCV, Lasso\n",
"from sklearn.model_selection import train_test_split\n",
"\n",
"# Strictly unnecessary, but cleans up some of Scikit-learn's annoying warnings...\n",
"import warnings\n",
"from sklearn.exceptions import ConvergenceWarning\n",
"warnings.filterwarnings(\"ignore\", category=ConvergenceWarning)"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {},
"outputs": [
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "d6e5469c311347a181f9e0b267e65f0e",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"HBox(children=(FloatProgress(value=0.0, max=361278.0), HTML(value='')))"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
}
],
"source": [
"# Creates a dataset to train against.\n",
"\n",
"vectorized_data = list()\n",
"bound_values = list()\n",
"\n",
"row_idxes, col_idxes = list(), list()\n",
"\n",
"nrows = 0\n",
"for ridx, row in enumerate(filter(lambda t: t[-1] == 'nM' and int(t[3]) > 0, tqdm(training_data))):\n",
" domain_tuple = tuple(row[:2])\n",
" dseq = alignment_mapping[domain_tuple]\n",
" pseq = align_phosphosite(row[2], 7, 7)\n",
" \n",
" # Vectorizing (using the same code as above) converts each sequence into a one-hot representation.\n",
" dseq_vec, pseq_vec = np.array(_vectorize_sequence(dseq, amino_acid_ordering)), np.array(_vectorize_sequence(pseq, amino_acid_ordering))\n",
" iseq_vec = np.array(_vectorize_interaction(dseq, pseq, amino_acid_ordering))\n",
" \n",
" # Sequences are laid out as [[domain sequence], [peptide sequence], [interacting sequences]]\n",
" # These values are constant for all iterations.\n",
" dlen, plen = len(dseq) * len(amino_acid_ordering), len(pseq) * len(amino_acid_ordering)\n",
" ilen = len(dseq) * len(pseq) * (len(amino_acid_ordering) ** 2)\n",
" total_size = dlen + plen + ilen\n",
" \n",
" # To work back to values - check what the outer offset (i.e. sequence type) is and then check the inner offset.\n",
" col_idxes.extend(k for k in dseq_vec)\n",
" col_idxes.extend(k + dlen for k in pseq_vec)\n",
" col_idxes.extend(k + (dlen + plen) for k in iseq_vec)\n",
" \n",
" row_idxes.extend([ridx] * (len(dseq_vec) + len(pseq_vec) + len(iseq_vec)))\n",
" \n",
" bound_values.append(float(row[-2]))\n",
" \n",
" ncols = total_size\n",
" nrows += 1\n",
"\n",
"values = np.ones(len(row_idxes))\n",
"\n",
"bound_values = np.array(bound_values)\n",
"vectorized_data_sp = sparse.coo_matrix((values, (row_idxes, col_idxes)), shape=(nrows, ncols))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Here, I am fitting a \"classic\" LASSO regressor to the data. In essence, this regressor matches HSM/ID (without the conversion into a likelihood model).\n",
"\n",
"Note: the regression introduces non-feasible $K_d$'s (i.e. negative values)"
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"(2601, 610827) (2601,)\n"
]
},
{
"data": {
"text/plain": [
"Lasso(alpha=1, copy_X=True, fit_intercept=True, max_iter=50, normalize=False,\n",
" positive=False, precompute=False, random_state=None, selection='cyclic',\n",
" tol=0.0001, warm_start=False)"
]
},
"execution_count": 14,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"print(vectorized_data_sp.shape, bound_values.shape)\n",
"\n",
"model = Lasso(alpha=1, max_iter=50)\n",
"model.fit(vectorized_data_sp, bound_values)"
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Correlation (Pearson): ⍴ = 0.888\n",
"Correlation (Spearman): ⍴ = 0.810\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 720x360 with 2 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"pred_regression = model.predict(vectorized_data_sp)\n",
"\n",
"corr_pearson, _ = stats.pearsonr(pred_regression, bound_values)\n",
"corr_spearman, _ = stats.spearmanr(pred_regression, bound_values)\n",
"\n",
"## Violates normality so Pearson isn't super accurate... Spearman is equivalent between two.\n",
"print(\"Correlation (Pearson): ⍴ = {0:.3f}\".format(corr_pearson))\n",
"print(\"Correlation (Spearman): ⍴ = {0:.3f}\".format(corr_spearman))\n",
" \n",
"fig, axes = plt.subplots(1,2, figsize=(10,5))\n",
"axes[0].scatter(pred_regression, bound_values, c='k')\n",
"sns.regplot(x=pred_regression, y=bound_values, order=1, color='k', line_kws={'color':\"#EA157A\"}, scatter_kws={'alpha':0.1}, ax=axes[1])\n",
"for ax in axes:\n",
" ax.set_ylim([-100, ax.get_ylim()[1]])\n",
" \n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This seemingly does much better than the probability model in terms of matching up $K_d$'s. The number of non-zero parameters, however, approaches the dataset size (2,601 data points) and, consequently, the regressor might just be \"memorizing\" the dataset."
]
},
{
"cell_type": "code",
"execution_count": 16,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"1118"
]
},
"execution_count": 16,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"np.count_nonzero(model.coef_)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"If we split the data into a train / test split, this problem comes into focus."
]
},
{
"cell_type": "code",
"execution_count": 17,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"Lasso(alpha=1, copy_X=True, fit_intercept=True, max_iter=50, normalize=False,\n",
" positive=False, precompute=False, random_state=None, selection='cyclic',\n",
" tol=0.0001, warm_start=False)"
]
},
"execution_count": 17,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Given the smaller size, this train / test split size probably doesn't make sense. Keeping to match papers.\n",
"X_train, X_test, ys_train, ys_test = train_test_split(vectorized_data_sp, bound_values, test_size=1/8)\n",
"\n",
"model = Lasso(alpha=1, max_iter=50)\n",
"model.fit(X_train, ys_train)"
]
},
{
"cell_type": "code",
"execution_count": 18,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Correlation (Pearson): ⍴ = 0.628\n",
"Correlation (Spearman): ⍴ = 0.449\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 720x360 with 2 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"pred_regression = model.predict(X_test)\n",
"\n",
"corr_pearson, _ = stats.pearsonr(pred_regression, ys_test)\n",
"corr_spearman, _ = stats.spearmanr(pred_regression, ys_test)\n",
"\n",
"## Violates normality so Pearson isn't super accurate... Spearman is equivalent between two.\n",
"print(\"Correlation (Pearson): ⍴ = {0:.3f}\".format(corr_pearson))\n",
"print(\"Correlation (Spearman): ⍴ = {0:.3f}\".format(corr_spearman))\n",
" \n",
"fig, axes = plt.subplots(1,2, figsize=(10,5))\n",
"axes[0].scatter(pred_regression, ys_test, c='k')\n",
"sns.regplot(x=pred_regression, y=ys_test, order=1, color='k', line_kws={'color':\"#EA157A\"}, scatter_kws={'alpha':0.1}, ax=axes[1])\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The mean absolute error ($\\log[K_d]$) shows the substantially worse performance of this as a predictor (compared with the likelihood model). "
]
},
{
"cell_type": "code",
"execution_count": 19,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"0.5114776832615514"
]
},
"execution_count": 19,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Need to convert negative predicted Kd's to \n",
"zeroed_pred_regression = np.array(pred_regression)\n",
"zeroed_pred_regression[zeroed_pred_regression <= 0] = 1e-5\n",
"\n",
"np.mean(np.abs(np.log10(zeroed_pred_regression) - np.log10(ys_test)))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The regressor model is fairly limited. I bring it up mostly for comparison. \n",
"\n",
"One potential way to use it is to fit domain-level models. In combination with structure-based masks (*e.g.* as used in AlQuraishi, *et. al.* (2014)), this may allow for detailed models of specific SH2 domains. Again, this a story of fewer data points and problems with generalizability in \"peptide space\". "
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"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": "3.6.10"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment