sample_motif_map.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import logomaker\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "original_grad_df = pd.read_csv('figures/fig1f_grad_seq101_model7_individual.csv')\n",
    "distilled_grad_df = pd.read_csv('figures/fig1f_grad_seq101_model7_distilled.csv')\n",
    "\n",
    "model_ix = 7\n",
    "seq_ix = 101\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "sample = distilled_grad_df.iloc[130:150,:].copy()\n",
    "# make noise smaller\n",
    "sample.iloc[:8,:] = sample.iloc[:8,:]*0.5\n",
    "sample.iloc[14:,:] = sample.iloc[14:,:]*0.5\n",
    "\n",
    "# make an \"unclean\" map\n",
    "# noisy_sample = sample + np.random.normal(0, 0.01, size=(sample.shape[0],sample.shape[1]))*0.5\n",
    "noisy_sample = distilled_grad_df.iloc[130:150,:].copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "noisy_sample.iloc[:8,:] = noisy_sample.iloc[:8,:] + (np.random.normal(0, 0.01, size=(sample.iloc[:8,:].shape[0],sample.iloc[:8,:].shape[1])))\n",
    "noisy_sample.iloc[14:,:] = noisy_sample.iloc[14:,:] + (np.random.normal(0, 0.01, size=(sample.iloc[14:,:].shape[0],sample.iloc[14:,:].shape[1])))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Axes: >"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x100 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "fig, ax = plt.subplots(figsize=(8,1))\n",
    "sns.heatmap(sample.transpose(), cmap='RdBu_r', center=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Axes: >"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x100 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "fig, ax = plt.subplots(figsize=(8,1))\n",
    "sns.heatmap(noisy_sample.transpose(), cmap='RdBu_r', center=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x200 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(2,1, figsize=(8,2))\n",
    "\n",
    "logomaker.Logo(distilled_grad_df.iloc[130:150,:].abs(), ax=axs[0])\n",
    "axs[0].axis('off')\n",
    "sns.heatmap(sample.transpose(), cmap='RdBu_r', center=0, ax=axs[1], cbar=False)\n",
    "axs[1].set_xticks([])\n",
    "axs[1].set_yticks([])\n",
    "axs[1].axhline(y=0, color='k',linewidth=3)\n",
    "axs[1].axhline(y=sample.transpose().shape[0], color='k',linewidth=3)\n",
    "axs[1].axvline(x=0, color='k',linewidth=3)\n",
    "axs[1].axvline(x=sample.transpose().shape[1], color='k',linewidth=3)\n",
    "fig.tight_layout()\n",
    "fig.show()\n",
    "fig.savefig(\"../sample_motif_analysis_clean.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x200 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(2,1, figsize=(8,2))\n",
    "\n",
    "noisy_motif = noisy_sample.abs()\n",
    "# mask\n",
    "m = np.zeros_like((noisy_motif*2).values)\n",
    "m[np.arange(len((noisy_motif*2))), (noisy_motif*2).values.argmax(1)] = 1\n",
    "noisy_motif_masked = noisy_motif*2*m\n",
    "logomaker.Logo(noisy_motif_masked, ax=axs[0])\n",
    "axs[0].axis('off')\n",
    "sns.heatmap(noisy_sample.transpose(), cmap='RdBu_r', center=0, ax=axs[1], cbar=False)\n",
    "axs[1].set_xticks([])\n",
    "axs[1].set_yticks([])\n",
    "axs[1].axhline(y=0, color='k',linewidth=3)\n",
    "axs[1].axhline(y=noisy_sample.transpose().shape[0], color='k',linewidth=3)\n",
    "axs[1].axvline(x=0, color='k',linewidth=3)\n",
    "axs[1].axvline(x=noisy_sample.transpose().shape[1], color='k',linewidth=3)\n",
    "fig.tight_layout()\n",
    "fig.show()\n",
    "fig.savefig(\"../sample_motif_analysis_noisy.pdf\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tf2",
   "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": "3.9.18"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}