figure3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A selection signature across datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import pyrepseq as prs\n",
    "import pyrepseq.plotting as pp\n",
    "\n",
    "from lib import *\n",
    "\n",
    "plt.style.use('tcrspace.mplstyle')\n",
    "\n",
    "pcDelta_kwargs = dict(pseudocount=0.0, normalize=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data loading and preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "back_norms, bins = prs.load_pcDelta_background()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dash 2017"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "dash = pd.read_csv(data_directory+'dash_human.csv')\n",
    "dash_epitopes = dash['epitope'].unique()\n",
    "chains = ['alpha', 'beta']\n",
    "cdr3_columns = {'alpha': 'cdr3_a_aa',\n",
    "                'beta' : 'cdr3_b_aa'}\n",
    "dash_pcDeltas_dict = {}\n",
    "for chain in chains:\n",
    "    dash_pcDeltas_dict[chain] = prs.pcDelta_grouped(dash, 'epitope',\n",
    "                                                    cdr3_columns[chain],\n",
    "                                                    **pcDelta_kwargs)\n",
    "dash_pcDeltas_dict['both'] = prs.pcDelta_grouped(dash, 'epitope',\n",
    "                                                [cdr3_columns[chain] for chain in chains],\n",
    "                                                **pcDelta_kwargs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Nolan 2020"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "mira = pd.read_csv(data_directory + 'mira/processed/peptide-detail-ci.csv')\n",
    "mira['nAminoAcids'] = mira['Amino Acids'].apply(lambda s: len(s.split(',')))\n",
    "mask = mira['nAminoAcids']==1\n",
    "mask &= mira['amino_acid'].apply(prs.isvalidcdr3)\n",
    "mira = mira[mask]\n",
    "mira = mira.groupby('Amino Acids').filter(lambda x: len(x) > 150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "mira_pcDeltas = prs.pcDelta_grouped(mira, 'Amino Acids', 'amino_acid', **pcDelta_kwargs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Minervina 2021"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "alpha 4 M 306\n",
      "alpha 4 W 204\n",
      "alpha 4 both 510\n",
      "alpha 8 M 300\n",
      "alpha 8 W 83\n",
      "alpha 8 both 383\n",
      "beta 4 M 325\n",
      "beta 4 W 218\n",
      "beta 4 both 543\n",
      "beta 8 M 287\n",
      "beta 8 W 70\n",
      "beta 8 both 357\n"
     ]
    }
   ],
   "source": [
    "minervina_hist_norms = {}\n",
    "# contracting from day 15 to 85\n",
    "chain_to_table = {'alpha' : 4,\n",
    "                  'beta': 3}\n",
    "for chain in ['alpha', 'beta']:\n",
    "    minervina = pd.read_csv(f'{data_directory}minervina/Table_S{chain_to_table[chain]}.tsv', sep='\\t')\n",
    "    for cd in [4, 8]:\n",
    "        for donor in ['M', 'W', 'both']:\n",
    "            if donor == 'both':\n",
    "                distances = []\n",
    "                for d in ['M', 'W']:\n",
    "                    seqs = minervina[(minervina['CD']==f'CD{cd}')]['CDR3.amino.acid.sequence']\n",
    "                    distances.append(prs.pdist(seqs))\n",
    "                np.concatenate(distances)\n",
    "            else:\n",
    "                seqs = minervina[(minervina['CD']==f'CD{cd}')\n",
    "                                &(minervina['donor']==donor)]['CDR3.amino.acid.sequence']\n",
    "                distances = prs.pdist(seqs)\n",
    "\n",
    "            print(chain, cd, donor, len(seqs))\n",
    "            hist, bins_ = np.histogram(distances, bins=bins)\n",
    "            minervina_hist_norm = hist/np.sum(hist)\n",
    "            minervina_hist_norms[chain, cd, donor] = minervina_hist_norm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tanno 2020"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "dfs = {filename : pd.read_csv(f'{data_directory}tanno/pruned/pdist_{filename}.csv')\n",
    "       for filename in tannofiles}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def N_from_pairs(pairs):\n",
    "    return round((1+np.sqrt(1+8*pairs))/2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "fabs = []\n",
    "for filename, df in dfs.items():    \n",
    "    if 'naive' in filename or 'memory' in filename:\n",
    "        continue\n",
    "    cab = df['count_pair']\n",
    "    if N_from_pairs(np.sum(cab)) < 25000:\n",
    "        continue\n",
    "        \n",
    "    fab = df['freq_pair']\n",
    "    fabs.append(fab)\n",
    "\n",
    "fab = np.array(fabs).mean(axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Minervina 2022"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(f'{data_directory}/minervina2022.csv')\n",
    "df = df.groupby('epitope').filter(lambda x: len(x) > 150)\n",
    "minervina22_pcDeltas = prs.pcDelta_grouped(df, 'epitope', ['cdr3a', 'cdr3b'], **pcDelta_kwargs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Putting it all togeter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def func(Delta, A, Deltac, pcb):\n",
    "    B = 1.0 - np.sum(A*np.exp(-Delta/Deltac)*pcb)\n",
    "    print(B, A)\n",
    "    return A*np.exp(-Delta/Deltac) + B"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.827241234264301 491677146.2559957\n",
      "0.9930056362857353 19906189.86111158\n",
      "0.9999980092480584 5665.745696926808\n",
      "0.8526354317166165 4994.726385751065\n",
      "0.9994580838135556 18.36752963639483\n",
      "0.6908118173399532 25962.36122062529\n",
      "0.9911667904559083 741.7197350434778\n",
      "0.9975086781034891 209.19492600784525\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 254.88x324 with 8 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes_arr = plt.subplots(figsize=(3.54, 4.5),\n",
    "                         ncols=3, nrows=3,\n",
    "                         sharex=False, sharey=False)\n",
    "\n",
    "\n",
    "title_kwargs = dict(ha='right', va='top', size='small')\n",
    "\n",
    "data_color = 'C4'\n",
    "fit_color = 'C3'\n",
    "\n",
    "# xy coordinates for the title annotation of each subplot\n",
    "title_y = 1.15\n",
    "title_x = 0.95\n",
    "\n",
    "\n",
    "Deltac = 1/np.log(10)\n",
    "\n",
    "#### alpha beta ####\n",
    "\n",
    "xmax = 17\n",
    "\n",
    "axes = axes_arr[0, :]\n",
    "hist_back_norm = back_norms['both']\n",
    "\n",
    "\n",
    "hist_norm = np.mean(dash_pcDeltas_dict['both'], axis=0)\n",
    "axes[0].plot(bins[:xmax], hist_norm[:xmax]/hist_back_norm[:xmax], 'o', color=data_color)\n",
    "A = (hist_norm[:xmax]/hist_back_norm[:xmax])[0]\n",
    "axes[0].plot(bins[:xmax], func(bins[:xmax], A, Deltac, hist_back_norm[:xmax]),\n",
    "             color=fit_color)\n",
    "axes[0].text(title_x, title_y,\n",
    "             'Tetramer\\nBinding\\n\\nEBV\\nCMV\\nFlu',\n",
    "             transform=axes[0].transAxes, **title_kwargs)\n",
    "\n",
    "\n",
    "hist_norm = minervina22_pcDeltas.mean(axis=0)\n",
    "axes[1].plot(bins[:xmax], hist_norm[:xmax]/hist_back_norm[:xmax], 'o', color=data_color)\n",
    "A = (hist_norm[:xmax]/hist_back_norm[:xmax])[0]\n",
    "axes[1].plot(bins[:xmax], func(bins[:xmax], A, Deltac, hist_back_norm[:xmax]),\n",
    "             color=fit_color)\n",
    "axes[1].text(title_x, title_y,\n",
    "             'Dextramer\\nBinding\\n\\nSARS-CoV-2',\n",
    "             transform=axes[1].transAxes, **title_kwargs)\n",
    "\n",
    "hist_norm = fab\n",
    "axes[2].plot(bins[:xmax], hist_norm[:xmax]/hist_back_norm[:xmax],\n",
    "             'o', color=data_color)\n",
    "A = hist_norm[0]/hist_back_norm[0]\n",
    "axes[2].plot(bins[:xmax], func(bins[:xmax], A, Deltac,\n",
    "             hist_back_norm[:xmax]), color=fit_color)\n",
    "axes[2].text(title_x, title_y,\n",
    "             'Unsorted\\nPBMCs\\n\\nMixed',\n",
    "             transform=axes[2].transAxes, **title_kwargs)\n",
    "\n",
    "\n",
    "axes[0].set_ylabel(r'$p_C(\\Delta)\\,/\\,p_{C,back}(\\Delta)$')\n",
    "for ax in axes:\n",
    "    ax.set_yscale('log')\n",
    "    ax.set_xlim(-0.5, 17.5)\n",
    "    ax.set_ylim(2e-1, 2e9)\n",
    "    ax.set_yticks(10**np.arange(0, 10, 2))\n",
    "    ax.set_xticks(np.arange(0, 20, 5))\n",
    "    ax.set_xlabel(r'$\\Delta$ (CDR3$\\alpha\\beta$)')\n",
    "for ax in axes[1:]:\n",
    "    ax.set_yticklabels([])\n",
    "\n",
    "#### alpha ####\n",
    "\n",
    "axes = axes_arr[1, :]\n",
    "\n",
    "hist_norm = np.mean(dash_pcDeltas_dict['alpha'], axis=0)\n",
    "hist_back_norm = back_norms['alpha'][:len(hist_norm)]\n",
    "axes[0].plot(bins[:-1], hist_norm / hist_back_norm, 'o', label='Data', color=data_color)\n",
    "A = (hist_norm / hist_back_norm)[0]\n",
    "axes[0].plot(bins[:-1], func(bins[:-1], A, Deltac, hist_back_norm),\n",
    "             label=r'$\\frac{K \\, 10^{-\\Delta}+1}{Z}$', color=fit_color)\n",
    "axes[0].text(title_x, title_y,\n",
    "             'Tetramer\\nBinding\\n\\nEBV\\nCMV\\nFlu',\n",
    "             transform=axes[0].transAxes, **title_kwargs)\n",
    "\n",
    "chain = 'alpha'\n",
    "cd = 8\n",
    "donor = 'both'\n",
    "hist_norm = minervina_hist_norms[chain, cd, donor]\n",
    "axes[1].plot(bins[:-1], hist_norm / hist_back_norm, 'o', color=data_color)\n",
    "A = (hist_norm / hist_back_norm)[0]\n",
    "axes[1].plot(bins[:-1], func(bins[:-1], A, Deltac, hist_back_norm), color=fit_color)\n",
    "axes[1].text(title_x, title_y,\n",
    "             'Temporal\\nTracking\\n\\nSARS-CoV-2',\n",
    "             transform=axes[1].transAxes,  **title_kwargs)\n",
    "\n",
    "axes[0].set_ylabel(r'$p_C(\\Delta)\\,/\\,p_{C,back}(\\Delta)$')\n",
    "for ax in axes:\n",
    "    ax.set_yscale('log')\n",
    "    ax.set_xlabel(r'$\\Delta$ (CDR3$\\alpha$)')\n",
    "    ax.set_xlim(-1, 13.5)\n",
    "    ax.set_ylim(4e-1, 1.1e5)\n",
    "    ax.set_xticks(np.arange(0, 14, 5))\n",
    "    ax.set_yticks(10**np.arange(6))\n",
    "for ax in axes[1:]:\n",
    "    ax.set_yticklabels([])\n",
    "axes_arr[1, 2].remove()\n",
    "\n",
    "\n",
    "#### beta ####\n",
    "\n",
    "axes = axes_arr[2, :]\n",
    "\n",
    "hist_norm = np.mean(dash_pcDeltas_dict['beta'], axis=0)\n",
    "hist_back_norm = back_norms['beta'][:len(hist_norm)]\n",
    "axes[0].plot(bins[:-1], hist_norm / hist_back_norm, 'o', label='Data', color=data_color)\n",
    "A = (hist_norm / hist_back_norm)[0]\n",
    "axes[0].plot(bins[:-1], func(bins[:-1], A, Deltac, hist_back_norm),\n",
    "             label=r'Reference $(K \\, 10^{-\\Delta}+1)/Z$', color=fit_color)\n",
    "axes[0].text(title_x, title_y,\n",
    "             'Tetramer\\nBinding\\n\\nEBV\\nCMV\\nFlu',\n",
    "             transform=axes[0].transAxes, **title_kwargs)\n",
    "handles, labels = axes[0].get_legend_handles_labels()\n",
    "fig.legend(handles, labels, loc='center right', ncol=1, bbox_to_anchor=(0.95, 0.5))\n",
    "\n",
    "hist_norm = np.mean(mira_pcDeltas, axis=0)\n",
    "axes[2].plot(bins[:-1], hist_norm / hist_back_norm, 'o', color=data_color)\n",
    "A = (hist_norm / hist_back_norm)[0]\n",
    "axes[2].plot(bins[:-1], func(bins[:-1], A, Deltac, hist_back_norm), color=fit_color)\n",
    "axes[2].text(title_x, title_y,\n",
    "             'Peptide-Stim\\nActivation\\n\\nSARS-CoV-2',\n",
    "             transform=axes[2].transAxes,  **title_kwargs)\n",
    "\n",
    "chain = 'beta'\n",
    "cd = 8\n",
    "donor = 'both'\n",
    "hist_norm = minervina_hist_norms[chain, cd, donor]\n",
    "axes[1].plot(bins[:-1], hist_norm / hist_back_norm, 'o', color=data_color)\n",
    "A = (hist_norm / hist_back_norm)[0]\n",
    "axes[1].plot(bins[:-1], func(bins[:-1], A, Deltac, hist_back_norm), color=fit_color)\n",
    "axes[1].text(title_x, title_y,\n",
    "             'Temporal\\nTracking\\n\\nSARS-CoV-2',\n",
    "             transform=axes[1].transAxes,  **title_kwargs)\n",
    "\n",
    "axes[0].set_ylabel(r'$p_C(\\Delta)\\,/\\,p_{C,back}(\\Delta)$')\n",
    "for ax in axes:\n",
    "    ax.set_yscale('log')\n",
    "    ax.set_xlabel(r'$\\Delta$ (CDR3$\\beta$)')\n",
    "    ax.set_xlim(-1, 13.5)\n",
    "    ax.set_ylim(4e-1, 1.1e5)\n",
    "    ax.set_xticks(np.arange(0, 14, 5))\n",
    "    ax.set_yticks(10**np.arange(6))\n",
    "for ax in axes[1:]:\n",
    "    ax.set_yticklabels([])\n",
    "\n",
    "pad = 1.0\n",
    "for ax, row in zip(axes_arr[:,0], [r'$\\alpha\\beta$', r'$\\alpha$', r'$\\beta$',]):\n",
    "    ax.annotate(row, xy=(0, 0.5), xytext=(-ax.yaxis.labelpad - pad, 0),\n",
    "                xycoords=ax.yaxis.label, textcoords='offset points',\n",
    "                size='large', ha='right', va='center')\n",
    "\n",
    "\n",
    "pp.label_axes(fig, xy=(-0.07, 1.12))\n",
    "fig.tight_layout(h_pad=0.2, w_pad=0.2, pad=0.0)\n",
    "\n",
    "fig.savefig('figures/raw/figure3.svg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}