rod_contact_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Rod contact classification\n",
    "This notebook contains the code to reproduce the classification of contacts between rods and bipolar cells, including the cross validation and the plots shown in figure 6"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import scipy.linalg\n",
    "from scipy.stats import itemfreq\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "from scipy.io import loadmat\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from sklearn import cross_validation\n",
    "from sklearn import svm\n",
    "%matplotlib inline\n",
    "matplotlib.rc('font',**{'family':'sans-serif','sans-serif':['Arial']})\n",
    "matplotlib.rcParams.update({'mathtext.default': 'regular'})\n",
    "matplotlib.rcParams.update({'font.size': 14})\n",
    "sns.set_style(\"whitegrid\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "BC_ids=np.loadtxt('data/BC_IDs_new')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "BC_in_rod_area=np.loadtxt('data/BC_in_rod_area')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "all_contacts=pd.read_pickle('data/rod_contact_data')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "BC_excluded=np.array([691,709,827,836])\n",
    "rod_excluded=np.array([3309])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "contact_summary=all_contacts[np.in1d(all_contacts['cell'],BC_excluded,invert=True)&\\\n",
    "                             np.in1d(all_contacts['cell'],rod_excluded,invert=True)][['cell','rod','type']]\\\n",
    "                            .drop_duplicates().reset_index().drop('index',axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "contact_summary.loc[(contact_summary['type']<63),'group']='off'\n",
    "contact_summary.loc[(contact_summary['type']>62),'group']='on'\n",
    "contact_summary.loc[(contact_summary['type']==71),'group']='rbc'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(contact_summary.shape[0]):\n",
    "    contact_summary.loc[i,'area']=np.sum(all_contacts['area'][(all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i])])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(contact_summary.shape[0]):\n",
    "    contact_summary.loc[i,'excentricity']=np.min(all_contacts['excentricity'][(all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i])])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(contact_summary.shape[0]):\n",
    "    contact_summary.loc[i,'angle']=np.max(all_contacts['angle'][(all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i])])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(contact_summary.shape[0]):\n",
    "    contact_summary.loc[i,'branch_dist']=np.min(all_contacts['branch_dist'][(all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i])])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(contact_summary.shape[0]):\n",
    "    contact_summary.loc[i,'end_dist']=np.min(all_contacts['end_dist'][(all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i])])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(contact_summary.shape[0]):\n",
    "    if (contact_summary['group'][i]=='on'):\n",
    "        contact_summary.loc[i,'height']=all_contacts['height'][(all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i])\\\n",
    "                                                                   &(all_contacts['angle']==contact_summary['angle'][i])].values[0]\n",
    "    elif (contact_summary['group'][i]=='off'):\n",
    "        contact_summary.loc[i,'height']=np.mean(all_contacts['height'][(all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i])])\n",
    "    else:\n",
    "        contact_summary.loc[i,'height']=all_contacts['height'][(all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i])\\\n",
    "                                                                   &(all_contacts['excentricity']==contact_summary['excentricity'][i])].values[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(contact_summary.shape[0]):\n",
    "    contact_summary.loc[i,'count']=np.sum((all_contacts['cell']==contact_summary['cell'][i])&(all_contacts['rod']==contact_summary['rod'][i]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "contact_summary=contact_summary[contact_summary['area']>0].reset_index().drop('index',axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "contact_summary['area']/=1e6\n",
    "contact_summary['branch_dist']/=1e3\n",
    "contact_summary['branch_dist']=np.clip(contact_summary['branch_dist'],1e-2,1e2)\n",
    "contact_summary['end_dist']/=1e3\n",
    "contact_summary['end_dist']=np.clip(contact_summary['end_dist'],1e-2,1e2)\n",
    "contact_summary['excentricity']/=1e3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "contact_summary_log_norm=contact_summary.copy()\n",
    "contact_summary_log_norm['excentricity']=np.log(contact_summary_log_norm['excentricity'])\n",
    "contact_summary_log_norm['area']=np.log(contact_summary_log_norm['area'])\n",
    "contact_summary_log_norm['branch_dist']=np.log(contact_summary_log_norm['branch_dist'])\n",
    "for i in range(4,contact_summary.shape[1]):\n",
    "    contact_summary_log_norm.ix[:,i]-=np.mean(contact_summary_log_norm.ix[:,i])\n",
    "    contact_summary_log_norm.ix[:,i]/=np.std(contact_summary_log_norm.ix[:,i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "classified_contacts=np.loadtxt('data/rod_training_data.csv',delimiter=',',usecols=range(4))\n",
    "classified_contacts=pd.DataFrame(classified_contacts,columns=['cell','rod','type','class'])\n",
    "classified_contacts=classified_contacts[np.in1d(classified_contacts['cell'],BC_excluded,invert=True)].reset_index().drop('index',axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "classified_contacts.loc[:,'group']=''\n",
    "classified_contacts.loc[:,'area']=0\n",
    "classified_contacts.loc[:,'excentricity']=0\n",
    "classified_contacts.loc[:,'angle']=0\n",
    "classified_contacts.loc[:,'branch_dist']=0\n",
    "classified_contacts.loc[:,'end_dist']=0\n",
    "classified_contacts.loc[:,'height']=0\n",
    "classified_contacts.loc[:,'count']=0\n",
    "cc_log_norm=classified_contacts.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(classified_contacts.shape[0]):\n",
    "    try:\n",
    "        classified_contacts.loc[i,'group']=contact_summary['group'][(contact_summary['cell']==classified_contacts['cell'][i])&(contact_summary['rod']==classified_contacts['rod'][i])].values[0]\n",
    "        classified_contacts.loc[i,'area']=contact_summary['area'][(contact_summary['cell']==classified_contacts['cell'][i])&(contact_summary['rod']==classified_contacts['rod'][i])].values[0]\n",
    "        classified_contacts.loc[i,'excentricity']=contact_summary['excentricity'][(contact_summary['cell']==classified_contacts['cell'][i])&(contact_summary['rod']==classified_contacts['rod'][i])].values[0]\n",
    "        classified_contacts.loc[i,'angle']=contact_summary['angle'][(contact_summary['cell']==classified_contacts['cell'][i])&(contact_summary['rod']==classified_contacts['rod'][i])].values[0]\n",
    "        classified_contacts.loc[i,'branch_dist']=contact_summary['branch_dist'][(contact_summary['cell']==classified_contacts['cell'][i])&(contact_summary['rod']==classified_contacts['rod'][i])].values[0]\n",
    "        classified_contacts.loc[i,'end_dist']=contact_summary['end_dist'][(contact_summary['cell']==classified_contacts['cell'][i])&(contact_summary['rod']==classified_contacts['rod'][i])].values[0]\n",
    "        classified_contacts.loc[i,'height']=contact_summary['height'][(contact_summary['cell']==classified_contacts['cell'][i])&(contact_summary['rod']==classified_contacts['rod'][i])].values[0]\n",
    "        classified_contacts.loc[i,'count']=contact_summary['count'][(contact_summary['cell']==classified_contacts['cell'][i])&(contact_summary['rod']==classified_contacts['rod'][i])].values[0]\n",
    "    except IndexError:\n",
    "        classified_contacts=classified_contacts.drop(classified_contacts.index[i]).reset_index().drop('index',axis=1)\n",
    "        cc_log_norm=cc_log_norm.drop(cc_log_norm.index[i]).reset_index().drop('index',axis=1)\n",
    "for i in range(cc_log_norm.shape[0]):\n",
    "    try:\n",
    "        cc_log_norm.loc[i,'group']=contact_summary_log_norm['group'][(contact_summary_log_norm['cell']==cc_log_norm['cell'][i])&(contact_summary_log_norm['rod']==cc_log_norm['rod'][i])].values[0]\n",
    "        cc_log_norm.loc[i,'area']=contact_summary_log_norm['area'][(contact_summary_log_norm['cell']==cc_log_norm['cell'][i])&(contact_summary_log_norm['rod']==cc_log_norm['rod'][i])].values[0]\n",
    "        cc_log_norm.loc[i,'excentricity']=contact_summary_log_norm['excentricity'][(contact_summary_log_norm['cell']==cc_log_norm['cell'][i])&(contact_summary_log_norm['rod']==cc_log_norm['rod'][i])].values[0]\n",
    "        cc_log_norm.loc[i,'angle']=contact_summary_log_norm['angle'][(contact_summary_log_norm['cell']==cc_log_norm['cell'][i])&(contact_summary_log_norm['rod']==cc_log_norm['rod'][i])].values[0]\n",
    "        cc_log_norm.loc[i,'branch_dist']=contact_summary_log_norm['branch_dist'][(contact_summary_log_norm['cell']==cc_log_norm['cell'][i])&(contact_summary_log_norm['rod']==cc_log_norm['rod'][i])].values[0]\n",
    "        cc_log_norm.loc[i,'end_dist']=contact_summary_log_norm['end_dist'][(contact_summary_log_norm['cell']==cc_log_norm['cell'][i])&(contact_summary_log_norm['rod']==cc_log_norm['rod'][i])].values[0]\n",
    "        cc_log_norm.loc[i,'height']=contact_summary_log_norm['height'][(contact_summary_log_norm['cell']==cc_log_norm['cell'][i])&(contact_summary_log_norm['rod']==cc_log_norm['rod'][i])].values[0]\n",
    "        cc_log_norm.loc[i,'count']=contact_summary_log_norm['count'][(contact_summary_log_norm['cell']==cc_log_norm['cell'][i])&(contact_summary_log_norm['rod']==cc_log_norm['rod'][i])].values[0]\n",
    "    except IndexError:\n",
    "        cc_log_norm=cc_log_norm.drop(cc_log_norm.index[i]).reset_index().drop('index',axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Classification of RBC to rod contacts\n",
    "\n",
    "Results needed to get the distance between OFF-CBC to rod contact points and true RBC to rod contacts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "clf_svm_rbc=svm.SVC(gamma=1e-1,C=10)\n",
    "clf_svm_rbc.fit(cc_log_norm.ix[cc_log_norm['group']=='rbc',5:].as_matrix(),cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix())\n",
    "\n",
    "contact_summary.loc[:,'prediction']=0\n",
    "contact_summary.ix[contact_summary['group']=='rbc','prediction']=clf_svm_rbc.predict(contact_summary_log_norm.ix[contact_summary_log_norm['group']=='rbc',4:11])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "contact_summary_off=contact_summary[contact_summary['group']=='off'].copy().reset_index().drop('index',axis=1)\n",
    "contact_summary_off.loc[:,'rbc_contact_dist']=10000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(contact_summary_off.shape[0]):\n",
    "    rod=contact_summary_off.ix[i,'rod']\n",
    "    rod_rbc_contacts=contact_summary[(contact_summary['rod']==rod)&(contact_summary['type']==71)&(contact_summary['prediction']==1)][['cell','excentricity']].as_matrix()\n",
    "    if rod_rbc_contacts.shape[0]>0:\n",
    "        rod_rbc_contacts=np.hstack((rod_rbc_contacts,np.zeros((rod_rbc_contacts.shape[0],3))))\n",
    "        for j in range(rod_rbc_contacts.shape[0]):\n",
    "            rod_rbc_contacts[j,2:]=all_contacts[(all_contacts['rod']==rod)&(all_contacts['cell']==rod_rbc_contacts[j,0])&np.isclose(all_contacts['excentricity'],rod_rbc_contacts[j,1]*1e3)][['x_pos','y_pos','z_pos']]\n",
    "        rod_cbc_contacts=all_contacts[(all_contacts['rod']==rod)&(all_contacts['cell']==contact_summary_off.ix[i,'cell'])][['x_pos','y_pos','z_pos']].as_matrix()\n",
    "        min_dist=[]\n",
    "        for j in range(rod_rbc_contacts.shape[0]):\n",
    "            for k in range(rod_cbc_contacts.shape[0]):\n",
    "                min_dist.append(np.linalg.norm(rod_rbc_contacts[j,2:]-rod_cbc_contacts[k,:]))\n",
    "        contact_summary_off.ix[i,'rbc_contact_dist']=np.min(np.array(min_dist))\n",
    "contact_summary_off['rbc_contact_dist']/=1e3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "contact_summary_log_norm_off=contact_summary_log_norm[contact_summary_log_norm['group']=='off'].copy().reset_index().drop('index',axis=1)\n",
    "contact_summary_log_norm_off.loc[:,'rbc_contact_dist']=np.log(contact_summary_off['rbc_contact_dist'])\n",
    "contact_summary_log_norm_off['rbc_contact_dist']-=np.mean(contact_summary_log_norm_off['rbc_contact_dist'])\n",
    "contact_summary_log_norm_off['rbc_contact_dist']/=np.std(contact_summary_log_norm_off['rbc_contact_dist'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "classified_contacts_off=classified_contacts[classified_contacts['group']=='off'].copy().reset_index().drop('index',axis=1)\n",
    "classified_contacts_off.loc[:,'rbc_contact_dist']=0\n",
    "for i in range(classified_contacts_off.shape[0]):\n",
    "    classified_contacts_off.loc[i,'rbc_contact_dist']=contact_summary_off['rbc_contact_dist']\\\n",
    "    [(contact_summary_off['cell']==classified_contacts_off['cell'][i])&(contact_summary_off['rod']==classified_contacts_off['rod'][i])].values[0]\n",
    "cc_log_norm_off=cc_log_norm[cc_log_norm['group']=='off'].copy().reset_index().drop('index',axis=1)\n",
    "cc_log_norm_off.loc[:,'rbc_contact_dist']=0\n",
    "for i in range(cc_log_norm_off.shape[0]):\n",
    "    cc_log_norm_off.loc[i,'rbc_contact_dist']=contact_summary_log_norm_off['rbc_contact_dist']\\\n",
    "    [(contact_summary_log_norm_off['cell']==cc_log_norm_off['cell'][i])&(contact_summary_log_norm_off['rod']==cc_log_norm_off['rod'][i])].values[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "classified_contacts_off=classified_contacts_off[(classified_contacts_off['rbc_contact_dist']<10)].reset_index().drop('index',axis=1)\n",
    "cc_log_norm_off=cc_log_norm_off[(cc_log_norm_off['rbc_contact_dist']<np.log(10))].reset_index().drop('index',axis=1)\n",
    "contact_summary_off=contact_summary_off[contact_summary_off['rbc_contact_dist']<10].reset_index().drop('index',axis=1)\n",
    "contact_summary_log_norm_off=contact_summary_log_norm_off[contact_summary_log_norm_off['rbc_contact_dist']<np.log(10)].reset_index().drop('index',axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "rows_to_delete=[]\n",
    "for i in range(contact_summary[contact_summary['group']=='off'].shape[0]):\n",
    "    if contact_summary_off[(contact_summary_off['cell']==contact_summary[contact_summary['group']=='off'].iloc[i]['cell'])&\\\n",
    "                    (contact_summary_off['rod']==contact_summary[contact_summary['group']=='off'].iloc[i]['rod'])].shape[0]<1:\n",
    "        rows_to_delete.append(contact_summary[contact_summary['group']=='off'].index[i])\n",
    "contact_summary=contact_summary.drop(rows_to_delete).reset_index().drop('index',axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "cc_plots=classified_contacts.copy()\n",
    "for i in range(cc_plots.shape[0]):\n",
    "    cc_plots.loc[i,'min_height']=np.min(all_contacts['height'][(all_contacts['cell']==cc_plots['cell'][i])&(all_contacts['rod']==cc_plots['rod'][i])])\n",
    "for i in range(cc_plots.shape[0]):\n",
    "    cc_plots.loc[i,'mean_height']=np.mean(all_contacts['height'][(all_contacts['cell']==cc_plots['cell'][i])&(all_contacts['rod']==cc_plots['rod'][i])])\n",
    "for i in range(cc_plots.shape[0]):\n",
    "    cc_plots.loc[i,'excent_min_height']=all_contacts['excentricity'][(all_contacts['cell']==cc_plots['cell'][i])&(all_contacts['rod']==cc_plots['rod'][i])\\\n",
    "                                                                   &(all_contacts['height']==cc_plots['min_height'][i])].values[0]\n",
    "for i in range(cc_plots.shape[0]):\n",
    "    cc_plots.loc[i,'area_log']=np.log10(cc_plots.ix[i,'area'])\n",
    "for i in range(cc_plots.shape[0]):\n",
    "    cc_plots.loc[i,'end_dist_log']=np.log10(cc_plots.ix[i,'end_dist'])\n",
    "cc_plots['excent_min_height']/=1e3"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Figures for poster/paper"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAJAAAABoCAYAAAAeh3nMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAGE9JREFUeJztnWtYU1fWx/8nIRDCJVwDyG0c0FFUgs5QFVppn6rtQLVS\nFHnaQucirVpbr9jxMmq1wMw8Oh3HS6vT1j4d7eiIiiJUZx4+WChUBQSpIpeCQIgkEG4hCYTknPcD\nL2cIEAInF7Dk90Vzkuy9Qv5Ze++1116HoCiKghUrDGFNtAFWnm6sArJiFFYBWTEKq4CsGIVVQFaM\nwiogK0ZhcQHV19cjNjZW51pmZibWrVuHXbt2ITMz09ImWTECiwqotbUVGRkZ4PF4Otfv3r0Lb29v\nAEBYWNiY2tJoNBCJRNBoNCa308o4oCaAdevW6TwuKSmh+vr6KJlMRm3YsEHv+86fP0/FxsZSsbGx\nVExMDDVz5kyqsbHR3OZaGQWbCRKtzuPi4mKEhYXB0dERBEHofd/atWuxdu1aAIBIJMKLL75oVjut\nGGZCBDQgko8++gh/+MMf4Obmht27dwMA3nnnnYkwyQpDCGqoO3hKGPBAubm58PPzm2hzpizWZbwV\no7AKyIpRjFtAT+mIZ8VM6BWQRCJBdXU1EhMTUVNTg5qaGlRXVyMhIcGS9lmZ5OhdhdXV1eHq1auo\nq6vD559/Tl9fuXKlRQyz8nSgV0CLFi3CokWL0NDQgICAAEvaZOUpwmAc6Nq1a7hx4wa4XC4oigJB\nEMjIyLCEbWaHJEmwWNZ1hDEYFNB3332H69evW8IWi6NSqeDg4DDRZjzV6P35qVQqqFQq+Pv7o6ys\njH6sUqksaZ9ZUavVE23CuOjq6ppoE4ah1wO9/fbbIAgCFEXh8OHD9HWCIPDVV18x7rC+vh5btmzB\nlStX6GuFhYV0GkdCQgLmz5/PuP3x0NPTY5F+TEVnZyecnZ0n2gwd9Aron//8JwBAqVTSQiIIAmw2\nm3Fn+tI5zpw5g08++QQajQZbtmzBJ598wriP8aBQKCzSj6no7OyEv7//RJuhg8E50Nq1ayGVSjF9\n+nTU19eDy+XC1tYW77777riX9B4eHti+fTuSk5OHPcdms8Fms9HX16f3/RcuXMCFCxcAmGb46ezs\nNLoNS9LW1jbRJgzDoIB8fHzw9ddfw8nJCQqFAikpKTh8+DDeeustxjGhodFsOzs7aDQaaDQa2NnZ\n6X2fKdM5KIqCTCZj/H5Lo1arJ6W9BgXU2tpKp19QFIWWlhZwuVyQJMm406HpHElJSdizZw80Gg02\nbtzIuN2xQlEUtm7digcPHiA0NJTOhpzMdHZ2TkqPaVBAO3bswIYNG9DX1weCILBt2zZcuHABiYmJ\njDv9xz/+AQDYu3cvACA8PBzh4eGM22OCXC5HbW0tuFyuRftlSnt7O9rb2yfajGHoFVBTUxN8fX0h\nEAiwb98++jpBEFi8eLFFjDMXBEEgOTkZ3t7esLW1nWhzxkRrayu6urqgVqsnlc16BZSTk4Pk5GSd\nfbAB0tPTzWqUJSBJEra2tlAoFMNWhZMNqVSKb7/9Fo8ePYJUKp1UCXR6BTSwUkpPT0dJSQmePHmC\nWbNmTbplJFOamprofz09PRm309fXBxsbm1FzuY3F09MThw8fRnBwMFpaWiaVgAxuBP3pT3/CN998\ngzNnzqCyshI7duywhF1mQywWo6WlBXfu3AEANDQ0GNWeUqk069EiiqLwww8/wNnZGSKRCD/++KPZ\n+mKCQQE9fPgQe/bsgb29PaKjo9HS0mIJu8xGR0cHgoODkZOTg46ODqMF1NPTY9aItlarxYcffgip\nVApbW1s8fvzYbH0xweAqjMvloqCgACRJ4v79+3B0dLSEXWaDxWKhu7sbDx8+hK+vL1xcXKBSqWBv\nb8+oPYVCAZIk4eTkZGJL+6EoClqtFjweDyKRaNJtvxj0QOnp6bh16xZ4PB6uXbuG1NRUS9hlNjgc\nDr10f/z4Me7du4fm5mbG7dXU1EAmk0EqlZrKRB04HA727t0LmUyGwMBAiMXiSZVWbFBALi4uiImJ\nQXJyMpYvX476+npL2GU2AgIC6MBhdXU1lEol4wgvRVG4efMmli1bZtad8traWiQlJWHevHkQCATo\n7e01W1/jxeAQ9v7778PZ2RkuLi4A+mMolg76mRIOh4O4uDicOnUKJElCpVKho6ODUVsSiQQdHR14\n9tln0dTUhODgYBNb2w9JkggMDKQ3tlUq1aQJgBoUkFwux4kTJ4zuSCKR4M9//jNcXFwQFBSEN954\nA0B/ZY7r16/D09MTCxcuxKpVq4zuyxB8Ph8RERHIy8vDrFmzIJfLGbVz5coV/PDDD1i+fDlmzJhh\nYiv/x5MnT3Ds2DFERUVBKBQa5YFMHYjUK6Bbt24BAHg8Hk6fPo1f/OIX9HNRUVHj7ujChQtISkpC\nWFgY3n77bSQkJIDNZtOVObRa7ZgrcxiDSqWCWq3G7NmzsXDhQnR3dzP2QHK5HD09PaAoCiUlJZg2\nbZqJre3ns88+g52dHSoqKiAUChmFDQoLC+Hm5obr169jw4YNUKvV9KhiDHoFVF5eDgCYO3cu1Go1\n/RhgJqDW1lb4+PgAAJydnSGXy+Hi4oLVq1dj3rx56Orqwt69e3Hy5Em9bZginePBgwe4dOkS7Ozs\nsGTJEvD5fLS0tND5TmOFJEnMnDkTUVFRSE1NxfHjxxnZM5Z+gH4vFBkZCbFYzKgdoVAIBwcHxMfH\n4/PPP0d2drZ5BbRp0yajGx/MtGnT0NzcDC8vL3R1ddGZdWOtzAGYJp1DIpFAoVDg8ePHoCgKMTEx\nUKlUaG1tHVdEurOzEzweD9nZ2fD390djYyN6e3tHTUdhAovFAkmScHR0REZGBvbs2cPoIEBZWRke\nPXqEzZs3Y/HixSbb2bdYdY7Vq1cjPT0djo6OWLZsGdLS0iakModIJALQn8BWWVmJ9vZ2vPnmm6is\nrByXgGQyGdRqNaKjowEAkZGRkMlkJh/GSJLE2rVr8de//hUhISGQSCTgcDjjbmdgA/z48ePw8PAw\nWWqsQQE9fvwYP/vZz+jHlZWVOvOhseLh4YEjR44Mu/7aa6/htddeG3d7TGhra6NjPgPLboVCAYIg\nUFpaisjIyDEPY1988QVaW1vR3NyMF154ARUVFQgICDC5gFgsFoKDgxEcHAwulws/Pz9GAqIoCmfO\nnEFvby+eeeYZhIaGmuRIk14B3b17F1VVVfjqq6+QlJREXz9//jyysrKM7ngiuH37NkiShLOzM3g8\nHpRKJX71q1+BxWJBJpPhxx9/HNNSXKFQwMXFBX/5y18QGRmJbdu2ISMjAy0tLdBoNLCxMa1j9/Pz\nw4oVK+gAIpNhkiAInDhxAvfu3QOfz0dqaqpJkvf0SnDatGng8XhwdHSEg4MDeDweHBwckJaWZnSn\nE4FKpcK9e/fAYrEQEhICpVIJgUCAJ0+e0K8pLCwcU1tisRiurq5ISUnBd999hx07dqCpqQkkSUIi\nkZjU7vv37yMvLw/ff/897THGuwzv6enB6dOnUVdXB4qi0NHRgezsbNTW1kKr1Rpln14B+fr6IjY2\nFrt374ZCoUBsbCyKi4uNSmWdSIqKitDX1weKolBcXAxnZ2dIpVJotVp62KqtrTW4rUGSJJqamjBt\n2jScPXsWbm5uUKvVmDVrFgCgsbHRpFsNQUFB2L9/Py0aFos15pMxSqUSDx8+RF5eHsRisU5G46NH\nj1BTU4O8vDzU1dWNephhNAwOgmlpaVi2bBkAYOvWrU9lMtlAnAboF0BzczM9B6qvr9f5wouLi0dt\nSywWQ61Wo6KiAq2trZDJZHBzc6Of7+rqMmnqqVgsxo0bN+g+xjI8dnd34/79+8jPz4dIJAJFUTh1\n6pTOawYcgVqtRnV1NfLy8lBdXT3u8IhBAREEAXd3dwD9EdzJtJE3Vpqbm+lgIUVROp/B2dlZZ+Jc\nUVGh9zNqNBrU1tYC6F/GDyzbBy8ygP49NlP9nUiSxP79+3H37l00NTUZnPjW1dWhoKBgmCft7u7W\neTw0GKnRaFBXV4f8/Pxx7Q0alPObb76JNWvWQCAQQCqV4ve///2YG58sDA6+kSQJPp9P/0GHpmEo\nFArI5fIRl7k1NTV0OgVJkuBwOGCz2cPa6OzsRENDAwIDA422nc/no6ysDEqlEjNmzDA4ydfn/dzc\n3HRExOfzR3ydRqOBXC6nnYYhDApo1apVWLVqFdra2uDq6jrp8lHGglKppP9PEITOxFnf64cKSCwW\n6ySfeXl5wcnJCfb29ujt7R2WJ1VVVQUnJyed4Y0JNjY2WLFiBYqLi9Hd3W3Qs82fPx/Nzc1oaGjQ\nCRYOFczQcAVBEBAIBAgMDBxXhNqggI4fP47//ve/6OnpgVKpREBAAM6dOzfmDiYDg5PmpVKpzpcw\nUrGIoUn2ra2tePDggc41qVSK9vZ2dHZ2oqioCC+99JLO8xRFobS0FOHh4UYlm3l4eGDPnj24fPky\n3e5oEAQBHx8f+Pj4QC6Xo6GhAWKxGLNnz4ZMJqO9cWhoKID+7ISAgAD4+fkxCg8YnAPl5+fj8uXL\nWLhwITIzM+n9rKcJX19f+v+enp46+15DP4+Tk5POF97d3Y2ysrJhX1x7eztIkoRGo8G333474oas\nRqNBSUmJ0fk7g/sez9zKyckJc+bMwaJFi7B48WJaPPb29vDy8oK7uzueffZZBAUFMd6CMSggLpcL\nrVYLhUIBd3d3xjm5EokE27Ztw8GDB3U8WGFhIT744AN88MEHuHfvHqO2DeHl5UUPJe3t7VixYgVd\nKGLoefOQkBAd915VVTVirGSw8IqKivDpp5+O2Hdvby/q6uqMsr+kpASFhYW4c+cOoxXewFDL4/FA\nEAQcHBzw61//GmFhYYyi2oMxKKDf/OY3uHjxIl588UW89NJLCAkJYdTRQDrHvn37cOvWLfpLOXPm\nDNLS0nDw4EGcPn2aUduGIAgCzzzzDABAIBCgsrISHA4HWq122GRxaLKcPu9RV1enM9SNluBlrAeS\ny+W4efMmXSmOCevWrUNcXBx8fHzg5uaGkJAQoyqtDGBwDqRSqejkr+joaFy9epVRR/rSOSiKGlNl\nDsC4dI4FCxYgPz8fcrkcUqmU/lIHJ9PPnTt3mKCCg4NH9IwBAQH05JzFYmH27Nkj9stmszF9+vRx\n2ToYkiQhEong6ekJFoulsyAYDywWC76+vhCLxQgODkZjY6PO0M6UUU+m3rhxA0VFRbhx4waA/vG3\nsbERr7766rg70pfOweVyx1SZAzAunYPD4eC5556j82A6OjrA4/HoL5cgCDz//PPD3ufp6YkFCxag\nvLxcR+AtLS0IDAxEe3s7hELhiEWx7O3tIRQKjdr5lsvlKC0thbe3N1555RXGX/r+/ftRXV2NoKAg\nSCQSeHl5MbZpMHoFtHTpUgiFQpw9exaJiYn0xJPpslRfOoclK3P88pe/REFBAezt7eHj4wN3d3c6\nwX7+/Pl6Yx8eHh6IiIhARUUFffoiNDQUL7/8MpqamhAQEDBsOPDz88PMmTON3lilKAr19fX0qs/V\n1ZVRO+Hh4Th06BCSkpKwadMmxseYhqL309na2sLX1xevv/46srOzdcZxJslm+tI5LFmZg81mY8mS\nJTh69CiePHmC559/HjY2NmCxWFiyZMmo77Wzs4NQKIREIsHDhw9BkiRWrFiB48eP64iHy+Vi7ty5\nRsd/BuDz+VizZg1OnDiBTz/9VKfc4HhQq9UoKChARkYGfHx8TFbSxuDPY+vWrYiPj4dAIDBJhxNN\naGgobGxsYG9vT1doFQqFeiOzgyEIAt7e3uDz+cjPz8fOnTvR2NhIJ9S7ublBKBQavbIZ2ud7772H\n6upqyGQyxgcA4uLiAACBgYEmEzcwBgG5uLggPj7eZB1ONGw2G2FhYfDx8YFIJIKfnx8WLVo0rjbs\n7e2xa9cu3L59m57Uurq6YsGCBWapOy0QCBATE4Pa2lp6DsgUNzc3kx4JMiggDoeD5ORkBAcH0/GR\nnTt3mswAS0MQBD777DO4uLggLi4Ofn5+jLwrh8PBnDlzcPfuXRAEgblz55q1aLmnpydqa2vh4eFh\nVDumPk9mUEC//e1vAYCu1PpToKGhgS5bwzSuBfR7HQcHB9jb25tsUqoPDw8PsFgskw4/psCggAID\nA/H3v/8ddXV18PPzw/vvv28Ju8zKvHnzIJfL0dvba/RpUicnJ4sUqHJ3d4erq+ukuzWDQWt2796N\nVatW4csvv0RcXBx27dplCbvMjr+/P7hcrtFDQkFBAUQiESorK01k2ci4uroyXsKbE4MC6u3tRXh4\nOGxtbbFw4cKnNqV1KAKBAAKBwOjKYi0tLXjhhRfw85//3ESWjYyTk9OYVoqWxuAQ5u/vjyNHjiA0\nNBRlZWU/mVs/2dnZmSSzgCRJnD17FlVVVZgzZ44JLBsZNptttLc0BwYFlJaWhtzcXNTV1UEoFGLp\n0qWWsMsiGFMbcYDExETweDyL1JqebBNoYBQBKZVKHD16FJs3b8bSpUuRm5uLgoICREREMLpF0tWr\nV1FUVISenh5s2rRJJ91z/fr19Pi+ceNGixXyNMWtnmxtbS1WdncyVofTOwfatWsXZsyYQS9Po6Ki\nMGPGDMaT6MzMTBw6dAibN2/WKR3c19eHxsZG2NnZwd3d3aIVSE2RzmBnZ2fy8/D6mIzJfHo9UFtb\nG1avXv2/F9rYICEhYcynUv/9738jKyuLjh8NCNHb21unHJxWq0V6ejpCQ0Nx7tw55OTkICYmZsQ2\nTX2zFaZnoQZjSQ9kzlLCTBl1DjT0lpBarXbMJxnj4+N1tkDWr18PoL9MyeDIr0wmg0gkQmhoKFxc\nXEb9Uk15sxVg5Hzo8WJra2vyo8xPE3o/+RtvvIGNGzfSG6kikQiXL19mvBfz6quvYu/evVAoFNi5\ncyckEgkuXbqEd955B8eOHUN5eTkUCgV9/wxLMPSsFBPMHYGe7BDUKPsTNTU1yMnJQWtrKwQCAZYv\nX46ZM2da0j69DHig3NxcRvMmjUaDrKwsvPzyy1NeBMYwqoAmM8YKSCwWY+XKlYiIiMDRo0cn5fzi\naWBybaxYCK1WC5lMZpJJ9FRnSs7+2Gw2QkJCsH37dp3aR1bGz1M7hGk0GjQ3N8Pb23tKr4ImmqdW\nQIMZEJMV8zLSj/Un8dNtbm42OiZkxTAjLVisHsjKmBnJA/0kBGRl4piSy3grpsMqICtGYRWQFaOw\nCsiKUfwklvHGcOzYMbp0L4/Hw+uvv44tW7ZAKBQC6C84FRISgvT0dPoIc2RkJKMKJcZw5coVZGVl\nISAgAFqtFkqlEj4+Pnj06BFcXV0hk8noJMCLFy+iqKgIbDYbarUaBw4cMFs245RehWVmZqKvrw9r\n1qwB0H9Hm7/97W8IDAzEgQMH6NfduXMHZWVlSE5OniBL+wVkZ2dH39zlww8/hEQiwfr16xEaGoqb\nN2+ioaEBUVFROHv2LA4ePAgAKC0tBZvNRnNzM/Lz86FQKJCYmEj/QIxlSnug+/fv0ydvgf4iCywW\nC7dv36ZTd3/3u98BAL755hu6RvT27dsn5ITEv/71L/qAQ3R0NNzd3XHy5Enw+XyUlpbiyJEjqKmp\noQtoAqBv4ldaWgoOh4OYmJhhda2NYUp7oIyMDJAkSWdOVlRU4PDhw/D39zfogU6cOIHa2lrExcUh\nIiLC7LYO9kAXL15EZWUlXFxcsGTJEoSGhkIikSAlJQX79+/HyZMn6VI633//Pdra2jB9+nTweDzk\n5uaitbXVZPUNprQHiouLw8cff4yUlBS6TlBKSgrOnz9v8L3vvvuuBSwcmTVr1uDQoUPIzs5GWVkZ\nXF1doVAoEB8fj6CgIMyfPx/vvfceHB0d0dfXhwMHDiA/Px//+c9/wOfz8dxzz5nMlintgawYj3UZ\nb8UopuwQptVqceTIEXR1daG3txdeXl6QyWRQKpWwtbVFe3s70tLSIBAIcOrUKfoeq7a2tti3b581\nB+n/mbJD2Llz58Dn8/HKK68AAK5du4acnBykpqbC3d0dX3zxBby9vcHj8VBZWUnfz/XWrVuYPn06\n8vLyUF9fj5aWFmzbts1ip2knG1P2Z1RVVYV169bRj1euXInCwkKkpqaCzWajvLycvj324GXxwC3P\nm5qa4ODggCVLlkzKogeWYsp6oIGVVkJCAgDgyy+/RHl5OXbv3g13d3cUFxcjIyMD0dHRuHPnDrZv\n3w4AyMrKgo+PD+zs7MDj8XDp0iX4+vrSxdinGlNWQBqNBqmpqejq6gJBEPD19YVUKoVSqQSXy0Vn\nZyfeeustLF68GB9//DEaGhrA4XBgb2+PP/7xj/j6669RUVEBNpuN+Ph4HS81lZiyArJiGqzLeCtG\nYRWQFaOwCsiKUVgFZMUorAKyYhRWAVkxiv8DBwEXM41ikEMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe1acbcd320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(4/2.54,3/2.54))\n",
    "sns.set(context=\"paper\",style='white',font='Arial',rc={\"xtick.major.size\": 0, \"ytick.major.size\": 4,\"lines.markeredgewidth\": 0.02,\"lines.markersize\":10})\n",
    "ax = sns.violinplot(x=\"group\",y=\"min_height\", data=cc_plots,hue='class',hue_order=[0,1],order=['off','rbc'],orient=\"v\",inner=None,\\\n",
    "                    linewidth=0,palette=sns.light_palette('gray',n_colors=3,reverse=True))\n",
    "ax2=sns.stripplot(x=\"group\", y=\"min_height\", data=cc_plots, hue='class',hue_order=[0,1],order=['off','rbc'],jitter=0.05,split=True, marker='x',size=2,linewidth=0.5,palette=['black','black'])\n",
    "ax.set(xlabel='',xticklabels=['OFF-\\nCBCs','RBCs'],ylim=(-0.7,1.5))\n",
    "plt.legend([])\n",
    "ax.set_ylabel('Contact height')\n",
    "sns.despine(offset=3)\n",
    "# plt.savefig('figures/cc_stats_min_height.svg',bbox_inches='tight',dpi=300)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAIkAAABoCAYAAADFPNPuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe1aa83da90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(4/2.54,3/2.54))\n",
    "sns.set(context=\"paper\",style='white',font='Arial',rc={\"xtick.major.size\": 0, \"ytick.major.size\": 4,\"lines.markeredgewidth\": 0.02,\"lines.markersize\":10})\n",
    "ax = sns.violinplot(x=\"group\",y=\"excentricity\", data=cc_plots,hue='class',hue_order=[0,1],order=['off','rbc'],orient=\"v\",inner=None,\\\n",
    "                    linewidth=0,palette=sns.light_palette('gray',n_colors=3,reverse=True))\n",
    "ax = sns.stripplot(x=\"group\", y=\"excentricity\", data=cc_plots, hue='class',hue_order=[0,1],order=['off','rbc'],jitter=0.05,split=True,\\\n",
    "                   size=2,marker='x',linewidth=0.5,palette=['black','black'])\n",
    "ax.set(xlabel='',xticklabels=['OFF-\\nCBCs','RBCs'],ylim=(0,3),yticks=[0,1,2,3])\n",
    "plt.legend([])\n",
    "ax.set_ylabel('Eccentricity [$\\mu m$]')\n",
    "sns.despine(offset=3)\n",
    "# plt.savefig('figures/cc_stats_eccentricity_min.svg',bbox_inches='tight',dpi=300)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAJYAAABrCAYAAACVDXslAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe1aa83f278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(4/2.54,3/2.54))\n",
    "sns.set(context=\"paper\",style='white',font='Arial',rc={\"xtick.major.size\": 0, \"ytick.major.size\": 4,\"lines.markeredgewidth\": 0.02,\"lines.markersize\":10})\n",
    "ax = sns.stripplot(x=\"group\", y=\"area\", data=cc_plots, hue='class',hue_order=[0,1],order=['off','rbc'],jitter=0, size=0,marker='x',palette=['black','black'])\n",
    "plt.legend([],loc='upper left')\n",
    "ax2 = ax.twinx()\n",
    "sns.violinplot(x=\"group\",y=\"area_log\", data=cc_plots,hue='class',hue_order=[0,1],order=['off','rbc'],orient=\"v\",inner=None,linewidth=0,palette=sns.light_palette('gray',n_colors=3,reverse=True),ax=ax2)\n",
    "sns.stripplot(x=\"group\", y=\"area_log\", data=cc_plots, hue='class',hue_order=[0,1],order=['off','rbc'],jitter=0.05,split=True,size=2,marker='x',linewidth=0.5,palette=['black','black',],ax=ax2)\n",
    "ax.set(yscale='log',ylim=(1e-4,1e2),xlabel='',xticklabels=['OFF-\\nCBCs','RBCs'],ylabel='Contact area [$\\mu m^2$]')\n",
    "ax.set_yticks([1e-4,1e-2,1,1e2])\n",
    "ax2.set(ylim=(-4,2),ylabel='',yticks=[])\n",
    "sns.despine(offset=3)\n",
    "plt.legend([])\n",
    "# plt.savefig('figures/cc_stats_area.svg',bbox_inches='tight',dpi=300)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Testing classification methods"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0001 1.0 0.6125 0.0 1.0\n",
      "0.0001 10.0 0.6125 0.0 1.0\n",
      "0.0001 100.0 0.6375 0.0 0.935483870968\n",
      "0.0001 1000.0 0.7875 0.183673469388 0.258064516129\n",
      "0.0001 10000.0 0.7125 0.224489795918 0.387096774194\n",
      "0.0001 100000.0 0.6875 0.265306122449 0.387096774194\n",
      "0.0001 1000000.0 0.725 0.224489795918 0.354838709677\n",
      "0.001 1.0 0.6125 0.0 1.0\n",
      "0.001 10.0 0.6375 0.0 0.935483870968\n",
      "0.001 100.0 0.8 0.183673469388 0.225806451613\n",
      "0.001 1000.0 0.7 0.224489795918 0.41935483871\n",
      "0.001 10000.0 0.7375 0.204081632653 0.354838709677\n",
      "0.001 100000.0 0.725 0.244897959184 0.322580645161\n",
      "0.001 1000000.0 0.675 0.30612244898 0.354838709677\n",
      "0.01 1.0 0.6375 0.0 0.935483870968\n",
      "0.01 10.0 0.7375 0.244897959184 0.290322580645\n",
      "0.01 100.0 0.725 0.224489795918 0.354838709677\n",
      "0.01 1000.0 0.7375 0.224489795918 0.322580645161\n",
      "0.01 10000.0 0.6875 0.30612244898 0.322580645161\n",
      "0.01 100000.0 0.6625 0.30612244898 0.387096774194\n",
      "0.01 1000000.0 0.6625 0.30612244898 0.387096774194\n",
      "0.1 1.0 0.7625 0.204081632653 0.290322580645\n",
      "0.1 10.0 0.725 0.265306122449 0.290322580645\n",
      "0.1 100.0 0.7125 0.265306122449 0.322580645161\n",
      "0.1 1000.0 0.7125 0.244897959184 0.354838709677\n",
      "0.1 10000.0 0.7125 0.244897959184 0.354838709677\n",
      "0.1 100000.0 0.7125 0.244897959184 0.354838709677\n",
      "0.1 1000000.0 0.7125 0.244897959184 0.354838709677\n"
     ]
    }
   ],
   "source": [
    "for gamma in np.logspace(-4,-1,4):\n",
    "    for C in np.logspace(0,6,7):\n",
    "        loo=cross_validation.LeaveOneOut(cc_log_norm_off.shape[0])\n",
    "        clf_svm=svm.SVC(gamma=gamma,C=C)\n",
    "        scores_svm_loo = cross_validation.cross_val_score(clf_svm,cc_log_norm_off.ix[:,5:].as_matrix(),\\\n",
    "                                                          cc_log_norm_off['class'].as_matrix(),cv=loo)\n",
    "        prediction_svm_loo = cross_validation.cross_val_predict(clf_svm,cc_log_norm_off.ix[:,5:].as_matrix(),\\\n",
    "                                                                cc_log_norm_off['class'].as_matrix(),cv=loo)\n",
    "        print(gamma,C,np.mean(scores_svm_loo),\\\n",
    "              np.sum(prediction_svm_loo>cc_log_norm_off['class'].as_matrix())/np.sum(cc_log_norm_off['class'].as_matrix()==0)\\\n",
    "              ,np.sum(prediction_svm_loo<cc_log_norm_off['class'].as_matrix())/np.sum(cc_log_norm_off['class'].as_matrix()==1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0001 1.0 0.787878787879 1.0 0.0\n",
      "0.0001 10.0 0.787878787879 1.0 0.0\n",
      "0.0001 100.0 0.878787878788 0.571428571429 0.0\n",
      "0.0001 1000.0 0.939393939394 0.238095238095 0.0128205128205\n",
      "0.0001 10000.0 0.949494949495 0.142857142857 0.025641025641\n",
      "0.0001 100000.0 0.949494949495 0.190476190476 0.0128205128205\n",
      "0.001 1.0 0.787878787879 1.0 0.0\n",
      "0.001 10.0 0.878787878788 0.571428571429 0.0\n",
      "0.001 100.0 0.939393939394 0.238095238095 0.0128205128205\n",
      "0.001 1000.0 0.949494949495 0.142857142857 0.025641025641\n",
      "0.001 10000.0 0.949494949495 0.190476190476 0.0128205128205\n",
      "0.001 100000.0 0.939393939394 0.142857142857 0.0384615384615\n",
      "0.01 1.0 0.878787878788 0.571428571429 0.0\n",
      "0.01 10.0 0.949494949495 0.190476190476 0.0128205128205\n",
      "0.01 100.0 0.939393939394 0.190476190476 0.025641025641\n",
      "0.01 1000.0 0.919191919192 0.190476190476 0.0512820512821\n",
      "0.01 10000.0 0.949494949495 0.142857142857 0.025641025641\n",
      "0.01 100000.0 0.949494949495 0.142857142857 0.025641025641\n",
      "0.1 1.0 0.929292929293 0.238095238095 0.025641025641\n",
      "0.1 10.0 0.949494949495 0.142857142857 0.025641025641\n",
      "0.1 100.0 0.939393939394 0.190476190476 0.025641025641\n",
      "0.1 1000.0 0.939393939394 0.190476190476 0.025641025641\n",
      "0.1 10000.0 0.939393939394 0.190476190476 0.025641025641\n",
      "0.1 100000.0 0.939393939394 0.190476190476 0.025641025641\n",
      "1.0 1.0 0.828282828283 0.809523809524 0.0\n",
      "1.0 10.0 0.848484848485 0.714285714286 0.0\n",
      "1.0 100.0 0.848484848485 0.714285714286 0.0\n",
      "1.0 1000.0 0.848484848485 0.714285714286 0.0\n",
      "1.0 10000.0 0.848484848485 0.714285714286 0.0\n",
      "1.0 100000.0 0.848484848485 0.714285714286 0.0\n"
     ]
    }
   ],
   "source": [
    "for gamma in np.logspace(-4,0,5):\n",
    "    for C in np.logspace(0,5,6):\n",
    "        loo=cross_validation.LeaveOneOut(cc_log_norm.ix[cc_log_norm['group']=='rbc',:].shape[0])\n",
    "        clf_svm=svm.SVC(gamma=gamma,C=C)\n",
    "        scores_svm_loo = cross_validation.cross_val_score(clf_svm,cc_log_norm.ix[cc_log_norm['group']=='rbc',5:].as_matrix(),\\\n",
    "                                                          cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix(),cv=loo)\n",
    "        prediction_svm_loo = cross_validation.cross_val_predict(clf_svm,cc_log_norm.ix[cc_log_norm['group']=='rbc',5:].as_matrix(),\\\n",
    "                                                                cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix(),cv=loo)\n",
    "        print(gamma,C,np.mean(scores_svm_loo),\\\n",
    "              np.sum(prediction_svm_loo>cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix())/np.sum(cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix()==0),\\\n",
    "              np.sum(prediction_svm_loo<cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix())/np.sum(cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix()==1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "off_svm_cv=svm.SVC(gamma=1e-3,C=100)\n",
    "rbc_svm_cv=svm.SVC(gamma=1e-1,C=10)\n",
    "loo=cross_validation.LeaveOneOut(cc_log_norm_off.shape[0])\n",
    "off_scores_svm_cv = cross_validation.cross_val_score(off_svm_cv,cc_log_norm_off.ix[:,5:].as_matrix(),\\\n",
    "                                                        cc_log_norm_off['class'].as_matrix(),cv=loo)\n",
    "off_prediction_cv = cross_validation.cross_val_predict(off_svm_cv,cc_log_norm_off.ix[:,5:].as_matrix(),\\\n",
    "                                                        cc_log_norm_off['class'].as_matrix(),cv=loo)\n",
    "loo=cross_validation.LeaveOneOut(cc_log_norm.ix[cc_log_norm['group']=='rbc',:].shape[0])\n",
    "rbc_scores_svm_cv = cross_validation.cross_val_score(rbc_svm_cv,cc_log_norm.ix[cc_log_norm['group']=='rbc',5:].as_matrix(),\\\n",
    "                                                        cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix(),cv=loo)\n",
    "rbc_prediction_cv = cross_validation.cross_val_predict(rbc_svm_cv,cc_log_norm.ix[cc_log_norm['group']=='rbc',5:].as_matrix(),\\\n",
    "                                                        cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix(),cv=loo)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "rbc_cv_plot=classified_contacts[classified_contacts['group']=='rbc'].copy().reset_index().drop('index',axis=1)\n",
    "rbc_cv_plot.loc[:,'prediction']=rbc_prediction_cv\n",
    "off_cv_plot=classified_contacts_off.copy()\n",
    "off_cv_plot.loc[:,'prediction']=off_prediction_cv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "cc_cv_plot=pd.concat((off_cv_plot[['cell','rod','type','class','group','prediction']],rbc_cv_plot[['cell','rod','type','class','group','prediction']]))\n",
    "cc_cv_plot.loc[:,'error']=np.abs(cc_cv_plot['class']-cc_cv_plot['prediction'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAALAAAACWCAYAAAB+QgJ1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe1aa853f60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(4/2.54,4/2.54))\n",
    "sns.set(font='Arial',style='white',context='paper',rc={\"xtick.major.size\": 4, \"ytick.major.size\": 4})\n",
    "ax=plt.scatter(off_cv_plot[(off_cv_plot['class']==1)&(off_cv_plot['prediction']==1)]['rbc_contact_dist'],\\\n",
    "               off_cv_plot[(off_cv_plot['class']==1)&(off_cv_plot['prediction']==1)]['area'],marker='o',c='k',s=20,lw=0.3,edgecolor='w').axes\n",
    "plt.scatter(off_cv_plot[(off_cv_plot['class']==0)&(off_cv_plot['prediction']==0)]['rbc_contact_dist'],\\\n",
    "               off_cv_plot[(off_cv_plot['class']==0)&(off_cv_plot['prediction']==0)]['area'],marker='^',c='k',s=20,lw=0.3,edgecolor='w')\n",
    "plt.scatter(off_cv_plot[(off_cv_plot['class']==1)&(off_cv_plot['prediction']==0)]['rbc_contact_dist'],\\\n",
    "               off_cv_plot[(off_cv_plot['class']==1)&(off_cv_plot['prediction']==0)]['area'],marker='o',c='w',s=17,lw=0.3)\n",
    "plt.scatter(off_cv_plot[(off_cv_plot['class']==0)&(off_cv_plot['prediction']==1)]['rbc_contact_dist'],\\\n",
    "               off_cv_plot[(off_cv_plot['class']==0)&(off_cv_plot['prediction']==1)]['area'],marker='^',c='w',s=17,lw=0.3)\n",
    "ax.set(xscale='log',xlim=(3e-1,5e0),yscale='log',ylim=(1e-3,1e1))\n",
    "ax.set(xticks=[5e-1,1e0,2e0,4e0])\n",
    "ax.set_xticklabels([0.5,1,2,4])\n",
    "plt.title('OFF-CBCs')\n",
    "ax.set_xlabel('Distance to RBC contact point [$\\mu m$]')\n",
    "ax.set_ylabel('Contact area [$\\mu m^2$]')\n",
    "sns.despine(offset=3)\n",
    "# plt.savefig('figures/classification_rod_off.svg',bbox_inches='tight',dpi=300,transparent=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAJYAAACcCAYAAACdmlKEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe1aa7f7198>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(4/2.54,4/2.54))\n",
    "sns.set(font='Arial',style='white',context='paper',rc={\"xtick.major.size\": 4, \"ytick.major.size\": 4})\n",
    "ax=plt.scatter(rbc_cv_plot[(rbc_cv_plot['class']==1)&(rbc_cv_plot['prediction']==1)]['excentricity'],\\\n",
    "               rbc_cv_plot[(rbc_cv_plot['class']==1)&(rbc_cv_plot['prediction']==1)]['area'],marker='o',c='k',s=20,lw=0.3,edgecolor='w').axes\n",
    "plt.scatter(rbc_cv_plot[(rbc_cv_plot['class']==0)&(rbc_cv_plot['prediction']==0)]['excentricity'],\\\n",
    "               rbc_cv_plot[(rbc_cv_plot['class']==0)&(rbc_cv_plot['prediction']==0)]['area'],marker='^',c='k',s=20,lw=0.3,edgecolor='w')\n",
    "plt.scatter(rbc_cv_plot[(rbc_cv_plot['class']==1)&(rbc_cv_plot['prediction']==0)]['excentricity'],\\\n",
    "               rbc_cv_plot[(rbc_cv_plot['class']==1)&(rbc_cv_plot['prediction']==0)]['area'],marker='o',c='w',s=17,lw=0.3)\n",
    "plt.scatter(rbc_cv_plot[(rbc_cv_plot['class']==0)&(rbc_cv_plot['prediction']==1)]['excentricity'],\\\n",
    "               rbc_cv_plot[(rbc_cv_plot['class']==0)&(rbc_cv_plot['prediction']==1)]['area'],marker='^',c='w',s=17,lw=0.3)\n",
    "ax.set(xscale='log',xlim=(2e-2,2e0),yscale='log',ylim=(1e-4,5e0))\n",
    "plt.title('RBCs')\n",
    "ax.set_xlabel('Eccentricity [$\\mu m$]')\n",
    "ax.set_ylabel('Contact area [$\\mu m^2$]')\n",
    "sns.despine(offset=3)\n",
    "# plt.savefig('figures/classification_rod_rbc.svg',bbox_inches='tight',dpi=300,transparent=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAIkAAAB0CAYAAACxKBEOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQV4iJIIkERJBkgiJIEmERJAkQiJI+h8Kr3ofbvG8uwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe1aa391470>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(4/2.54,4/2.54))\n",
    "sns.set(font='Arial',context='paper',style='white',rc={\"xtick.major.size\": 0, \"ytick.major.size\": 4})\n",
    "ax=sns.barplot(x='group',y='error',hue='class',data=cc_cv_plot,ci=None,palette=[np.array([0.8,0.8,0.8,1.]),np.array([0.4,0.4,0.4,1.])])\n",
    "ax.set(ylabel='error rate',xlabel='',xticklabels=['Off CBC','RBC'],yticks=[0,0.1,0.2,0.3])\n",
    "plt.legend(bbox_to_anchor=(1, 1.08),labels=['false negative','false positive'])\n",
    "sns.despine()\n",
    "# plt.savefig('figures/classication_performance_rods.svg',bbox_inches='tight',dpi=300,transparent=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(C=10, cache_size=200, class_weight=None, coef0=0.0,\n",
       "  decision_function_shape=None, degree=3, gamma=0.1, kernel='rbf',\n",
       "  max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
       "  tol=0.001, verbose=False)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf_svm_rbc=svm.SVC(gamma=1e-1,C=10)\n",
    "clf_svm_rbc.fit(cc_log_norm.ix[cc_log_norm['group']=='rbc',5:].as_matrix(),cc_log_norm.ix[cc_log_norm['group']=='rbc']['class'].as_matrix())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "contact_summary.loc[:,'prediction']=0\n",
    "contact_summary.ix[contact_summary['group']=='rbc','prediction']=clf_svm_rbc.predict(contact_summary_log_norm.ix[contact_summary_log_norm['group']=='rbc',4:11])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "clf_svm_off=svm.SVC(gamma=1e-3,C=100)\n",
    "clf_svm_off.fit(cc_log_norm_off.ix[:,5:].as_matrix(),cc_log_norm_off['class'].as_matrix())\n",
    "contact_summary.ix[contact_summary['group']=='off','prediction']=clf_svm_off.predict(contact_summary_log_norm_off.ix[:,4:].as_matrix())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# contact_summary.to_pickle('data/rod_contact_predictions')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "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": "3.4.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}