ipython.notebooks/Cell2015_N8_HAL_Genome_Predictions.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/net/gs/vol3/software/modules-sw-python/2.7.3/pandas/0.14.0/Linux/RHEL6/x86_64/lib/python2.7/site-packages/pandas/io/excel.py:626: UserWarning: Installed openpyxl is not supported at this time. Use >=1.6.1 and <2.0.0.\n",
      "  .format(openpyxl_compat.start_ver, openpyxl_compat.stop_ver))\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import scipy\n",
    "import scipy.sparse\n",
    "import scipy.stats\n",
    "import os\n",
    "import scipy.io as sio\n",
    "import dnatools\n",
    "%matplotlib inline\n",
    "from pylab import *\n",
    "\n",
    "# Plotting Params:\n",
    "rc('mathtext', default='regular')\n",
    "fsize=14\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Make directory to save results:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "resultsdir = '../results/N8_HAL_Genome_Predictions/'\n",
    "if not os.path.exists(resultsdir):\n",
    "    os.makedirs(resultsdir)\n",
    "figdir = '../figures/N8_HAL_Genome_Predictions/'\n",
    "if not os.path.exists(figdir):\n",
    "    os.makedirs(figdir)\n",
    "    \n",
    "#Choose if you want to actually save the plots:\n",
    "SAVEFIGS = True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load other splice site models:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "f = open('../ref/MAXENT_ALL_SD.txt')\n",
    "all_ss_scores = {'MAXENT':{},\n",
    "                 'MDD':{},\n",
    "                 'MM':{},\n",
    "                 'WMM':{}\n",
    "                 }\n",
    "while True:\n",
    "    line =  f.readline()[:-1]\n",
    "    line =  f.readline()[:-1]\n",
    "    if len(line)==0:\n",
    "        break\n",
    "    all_ss_scores['MAXENT'][line.split('\\t')[0]] = float(line.split('\\t')[1][8:])\n",
    "    all_ss_scores['MDD'][line.split('\\t')[0]] = float(line.split('\\t')[2][5:])\n",
    "    all_ss_scores['MM'][line.split('\\t')[0]] = float(line.split('\\t')[3][4:])\n",
    "    all_ss_scores['WMM'][line.split('\\t')[0]] = float(line.split('\\t')[4][5:])\n",
    "f.close()\n",
    "\n",
    "def score_ss(seq,ss_scores):\n",
    "    return -ss_scores[seq[198:198+9]]+ss_scores[seq[-203:-203+9]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load HAL:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "data = sio.loadmat('../results/N7_A5SS_Model_Predictions/model_full_data.mat')\n",
    "w = data['Mer_scores'][:4**6*8]\n",
    "mer6_dict = dict(zip(dnatools.make_mer_list(6),range(4**6)))\n",
    "w_mat = w.reshape(4**6,size(w)/4**6)\n",
    "logit = lambda x: log(x)-log(1-x)\n",
    "expit = lambda x: 1./(1.+exp(-x))\n",
    "\n",
    "def score_seq_pos(seq):\n",
    "    seq_scores = np.zeros(160)\n",
    "    for b in range(40):\n",
    "        seq_scores[b:b+6] += w_mat[mer6_dict[seq[b:b+6]],0]/6.\n",
    "    for b in range(40,70):\n",
    "        seq_scores[b:b+6] += w_mat[mer6_dict[seq[b:b+6]],1]/6.\n",
    "    for b in [77]:\n",
    "        seq_scores[b:b+6] += w_mat[mer6_dict[seq[b:b+6]],2]/6.\n",
    "    for b in [78]:\n",
    "        seq_scores[b:b+6] += w_mat[mer6_dict[seq[b:b+6]],3]/6.\n",
    "    for b in [79]:\n",
    "        seq_scores[b:b+6] += w_mat[mer6_dict[seq[b:b+6]],4]/6.\n",
    "    for b in [80]:\n",
    "        seq_scores[b:b+6] += w_mat[mer6_dict[seq[b:b+6]],5]/6.\n",
    "    for b in range(86,120-5):\n",
    "        seq_scores[b:b+6] += w_mat[mer6_dict[seq[b:b+6]],6]/6.\n",
    "    for b in range(115,155):\n",
    "        seq_scores[b:b+6] += w_mat[mer6_dict[seq[b:b+6]],7]/6.\n",
    "    return seq_scores\n",
    "\n",
    "def predict_prob(seq):\n",
    "    sd1_score = sum(score_seq_pos(seq[201-80:201+80]))\n",
    "    sd2_score = sum(score_seq_pos(seq[-200-80:-200+80]))\n",
    "    return sd2_score-sd1_score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load SNP data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "A5SS_data = pd.read_csv(resultsdir+'A5SS_Miso_events.csv',index_col=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Only include Events Found in 10 Samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "A5SS_data_filtered = A5SS_data[A5SS_data.sample_num>=10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "preds_comp = {}\n",
    "preds_comp['HAL'] = expit(A5SS_data_filtered['Seqs'].apply(predict_prob))\n",
    "for alg in all_ss_scores.keys():\n",
    "    preds_comp['$\\Delta$'+alg] = A5SS_data_filtered['Seqs'].apply(lambda s:score_ss(s,all_ss_scores[alg]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "prediction_df = pd.DataFrame(preds_comp).join(A5SS_data_filtered)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "prediction_df[[u'sample_num',\n",
    "                     u'miso_posterior_mean',\n",
    "                     u'ci_low',\n",
    "                     u'ci_high',\n",
    "                     u'HAL',\n",
    "                     u'$\\Delta$MAXENT',\n",
    "                     u'$\\Delta$MDD',\n",
    "                     u'$\\Delta$MM',\n",
    "                     u'$\\Delta$WMM',\n",
    "                    ]].to_csv(resultsdir+'A5SS_MISO_Event_Predictions.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/net/shendure/vol1/home/abros/VirtualEnvs/analysisenv/lib/python2.7/site-packages/ipykernel/__main__.py:4: DeprecationWarning: using a non-integer number instead of an integer will result in an error in the future\n"
     ]
    }
   ],
   "source": [
    "dists = A5SS_data_filtered.Seqs.apply(len).order()-400\n",
    "bin_edges = []\n",
    "for i in arange(0,1.0,0.2):\n",
    "    bin_edges.append(dists.values[np.round(i*len(dists))])\n",
    "bin_edges.append(5000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "dist_strs = []\n",
    "preds_r2 = []\n",
    "ss_compar_r2 = {}\n",
    "for alg in preds_comp.keys():\n",
    "    ss_compar_r2[alg]=[]\n",
    "\n",
    "\n",
    "for i in range(5):\n",
    "    dist_strs.append(str(bin_edges[i])+'-'+str(bin_edges[i+1]-1))\n",
    "    events = A5SS_data_filtered[((A5SS_data_filtered.Seqs.apply(len)-400)>=bin_edges[i]) & ((A5SS_data_filtered.Seqs.apply(len)-400)<bin_edges[i+1])].index\n",
    "    for alg in preds_comp.keys():\n",
    "        ss_compar_r2[alg].append(scipy.stats.pearsonr(preds_comp[alg][events],A5SS_data_filtered.miso_posterior_mean[events])[0]**2)\n",
    "    #preds_r2.append(scipy.stats.pearsonr(expit(preds[events]),A5SS_data_filtered.miso_posterior_mean[events])[0]**2)\n",
    "    \n",
    "#ss_compar_r2['HAL'] = preds_r2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAFRCAYAAABaC/vdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQND/134wAwAAALAENdsP5QKB+ZsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb650d10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = figure()\n",
    "ax = fig.add_subplot(111)\n",
    "dist_pred_df = pd.DataFrame(ss_compar_r2)\n",
    "dist_pred_df.index=dist_strs\n",
    "dist_pred_df.iloc[:,[4,0,1,2,3]].plot(ax=ax,kind='bar')\n",
    "ax.tick_params(labelsize=fsize)\n",
    "ax.set_xlabel('Distance Between Alternative Splice Donors (nt)',fontsize=fsize)\n",
    "ax.set_ylabel('$R^2$',fontsize=fsize)\n",
    "ax.set_title(\"Predicting 5' Splice Site Usage\\n in the Human Transcriptome (6152 Events)\",fontsize=14)\n",
    "ax.set_xticklabels(dist_pred_df.index.values,rotation=30)\n",
    "leg = ax.legend(fontsize=fsize)\n",
    "if SAVEFIGS:\n",
    "    figname = 'HAL_A5SS_comparisons_grouped_by_distance'\n",
    "    fig.savefig(figdir+figname+'.pdf',dpi=300,bbox_inches='tight')\n",
    "    fig.savefig(figdir+figname+'.eps',dpi=300,bbox_inches='tight')\n",
    "    fig.savefig(figdir+figname+'.png',dpi=300,bbox_inches='tight')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6152"
      ]
     },
     "execution_count": 133,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(A5SS_data_filtered)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}