Tutorial 2 - Fingerprints and Classification.ipynb

{
 "metadata": {
  "name": "Tutorial 2 - Fingerprints and Classification"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Fingerprints and Classification\n",
      "===="
     ]
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Helper methods"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.metrics.metrics import classification_report, confusion_matrix, roc_curve,auc,zero_one_score\n",
      "\n",
      "def confusion_matrix_summary(acts,preds):\n",
      "    from cStringIO import StringIO\n",
      "    file_str = StringIO()\n",
      "    vTab=confusion_matrix(acts,preds)\n",
      "    #print vTab\n",
      "    nResultCodes=len(vTab)\n",
      "    file_str.write('\\n\\tResults Table (experiment in rows):\\n')\n",
      "    colCounts = numpy.sum(vTab,0)\n",
      "    rowCounts = numpy.sum(vTab,1)\n",
      "    print\n",
      "    for i in range(nResultCodes):\n",
      "        if rowCounts[i]==0: rowCounts[i]=1\n",
      "        row = vTab[i]\n",
      "        file_str.write('    ')\n",
      "        for j in range(nResultCodes):\n",
      "            entry = row[j]\n",
      "            file_str.write(' % 6d'%entry),\n",
      "        file_str.write('     | % 4.2f\\n'%(100.*vTab[i,i]/rowCounts[i]))\n",
      "    file_str.write('    ')\n",
      "    for i in range(nResultCodes):\n",
      "        file_str.write('-------')\n",
      "    file_str.write('\\n')\n",
      "    file_str.write('    '),\n",
      "    for i in range(nResultCodes):\n",
      "        if colCounts[i]==0: colCounts[i]=1\n",
      "        file_str.write(' % 6.2f'%(100.*vTab[i,i]/colCounts[i])),\n",
      "    file_str.write('\\n')\n",
      "    return file_str.getvalue()\n",
      "\n",
      "def createROCPlot(pl,observations,probabilities,caption):\n",
      "    fpr, tpr, thresholds = roc_curve(observations, probabilities)\n",
      "    roc_auc = auc(fpr, tpr)\n",
      "    print \"Area under the ROC curve : %f\" % roc_auc\n",
      "    pl.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc)\n",
      "    pl.plot([0, 1], [0, 1], 'k--')\n",
      "    pl.set_xlim([0.0, 1.0])\n",
      "    pl.set_ylim([0.0, 1.0])\n",
      "    pl.set_xlabel('False Positive Rate')\n",
      "    pl.set_ylabel('True Positive Rate')\n",
      "    pl.set_title(caption)\n",
      "    \n",
      "def createPropAccPlot(pl,observations,probabilities,caption):\n",
      "    fpr, tpr, thresholds = roc_curve(observations, probabilities)\n",
      "    accuracies = []\n",
      "    for t in thresholds:\n",
      "        \n",
      "        predictions = [1 if p >= t else 0 for p in probabilities]\n",
      "        acc = zero_one_score(observations,predictions)\n",
      "        accuracies.append(acc)\n",
      "    pl.plot(thresholds, accuracies, label='Prob. vs Accuracy curve')\n",
      "    pl.plot([0, 1], [0, 1], 'k--')\n",
      "    pl.set_xlim([0.0, 1.0])\n",
      "    pl.set_ylim([0.0, 1.0])\n",
      "    pl.set_xlabel('Probability threshold')\n",
      "    pl.set_ylabel('Accuracy')\n",
      "    pl.set_title(caption)\n",
      "    "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 16
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Dataset"
     ]
    },
    {
     "cell_type": "heading",
     "level": 4,
     "metadata": {},
     "source": [
      "Publicly available binary Ames test (mutagenicity) data compiled in \n",
      "\n",
      "\"Benchmark Data Set for in Silico Prediction of Ames Mutagenicity\", Katja Hansen et al., JCIM, 2009, 49(9),pp 2077-2081"
     ]
    },
    {
     "cell_type": "heading",
     "level": 4,
     "metadata": {},
     "source": [
      "Data available from http://doc.ml.tu-berlin.de/toxbenchmark/smiles_cas_N6512.smi"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from rdkit import Chem\n",
      "import csv\n",
      "#obtained from http://doc.ml.tu-berlin.de/toxbenchmark/smiles_cas_N6512.smi\n",
      "with open('data/smiles_cas_N6512.smi','rU') as train_file:\n",
      "    r = csv.reader(train_file, delimiter='\\t')\n",
      "    r.next()\n",
      "    molecules = []\n",
      "    observations = []\n",
      "    for row in r:\n",
      "        try:\n",
      "         temp = Chem.MolFromSmiles(row[0])        \n",
      "        except e:\n",
      "            print e,row\n",
      "            continue\n",
      "        molecules.append(temp)\n",
      "        observations.append(int(row[2]))\n",
      "#molecules = [Chem.MolFromSmiles(d[0]) for d in row]\n",
      "#observations = [int(d[2]) for d in raw]\n",
      "data = [(x,y) for x,y in zip(molecules,observations) if x is not None]\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 17
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Fingerprint calculation"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from rdkit.Chem import AllChem\n",
      "import random\n",
      "\n",
      "ind = range(len(data))\n",
      "random.seed(0)\n",
      "random.shuffle(ind)\n",
      "half = int(round(0.5*len(ind)))\n",
      "train = ind[:half]\n",
      "test = ind[half:]\n",
      "\n",
      "train_morgan_fps = []\n",
      "train_rd_fps = []\n",
      "train_ys = []\n",
      "train_info=[]\n",
      "test_info=[]\n",
      "\n",
      "for i in train:\n",
      "    mol = data[i][0]\n",
      "    info = {}\n",
      "    temp = AllChem.GetMorganFingerprintAsBitVect(mol,2,2048,bitInfo=info)\n",
      "    train_morgan_fps.append(temp)\n",
      "    train_info.append(info)\n",
      "    train_rd_fps.append(Chem.RDKFingerprint(mol))\n",
      "    train_ys.append(int(data[i][1]))\n",
      "    \n",
      "test_morgan_fps = []\n",
      "test_rd_fps = []\n",
      "test_ys = []\n",
      "    \n",
      "for i in test:\n",
      "    mol = data[i][0]\n",
      "    info = {}\n",
      "    temp = AllChem.GetMorganFingerprintAsBitVect(mol,2,2048,bitInfo=info)\n",
      "    test_morgan_fps.append(temp)\n",
      "    test_info.append(info)\n",
      "    test_rd_fps.append(Chem.RDKFingerprint(mol))\n",
      "    test_ys.append(int(data[i][1]))\n",
      "    \n",
      "import cPickle\n",
      "outf=file('data/fps.pkl','wb+')\n",
      "cPickle.dump((train,train_morgan_fps,train_info,train_rd_fps,train_ys),outf)\n",
      "cPickle.dump((test,test_morgan_fps,test_info,test_rd_fps,test_ys),outf)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 18
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import cPickle\n",
      "inf=file('data/fps.pkl','rb+')\n",
      "(train,train_morgan_fps,train_info,train_rd_fps,train_ys)=cPickle.load(inf)\n",
      "(test,test_morgan_fps,test_info,test_rd_fps,test_ys)=cPickle.load(inf)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 19
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Binary Classification"
     ]
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "Naive Bayes"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.naive_bayes import BernoulliNB\n",
      "morgan_bnb = BernoulliNB(fit_prior=False)\n",
      "morgan_bnb.fit(train_morgan_fps,train_ys)\n",
      "morgan_predictions = morgan_bnb.predict(test_morgan_fps)\n",
      "print 'Morgan Fingerprints'\n",
      "print confusion_matrix_summary(test_ys,morgan_predictions)\n",
      "print classification_report(test_ys,morgan_predictions)\n",
      "\n",
      "\n",
      "print '------------------------------------------------------------'\n",
      "print 'RDKit Fingerprints'\n",
      "rd_bnb = BernoulliNB(fit_prior=False)\n",
      "rd_bnb.fit(train_rd_fps,train_ys)\n",
      "rd_predictions = rd_bnb.predict(test_rd_fps)\n",
      "print confusion_matrix_summary(test_ys,rd_predictions)\n",
      "print classification_report(test_ys,rd_predictions)\n",
      "\n",
      "morgan_probabilities = [p[1] for p in morgan_bnb.predict_proba(test_morgan_fps)]\n",
      "rd_probabilities = [p[1] for p in rd_bnb.predict_proba(test_rd_fps)]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Morgan Fingerprints\n",
        "\n",
        "\n",
        "\tResults Table (experiment in rows):\n",
        "       1132    368     |  75.47\n",
        "        530   1194     |  69.26\n",
        "    --------------\n",
        "      68.11  76.44\n",
        "\n",
        "             precision    recall  f1-score   support\n",
        "\n",
        "          0       0.68      0.75      0.72      1500\n",
        "          1       0.76      0.69      0.73      1724\n",
        "\n",
        "avg / total       0.73      0.72      0.72      3224\n",
        "\n",
        "------------------------------------------------------------\n",
        "RDKit Fingerprints\n",
        "\n"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "\tResults Table (experiment in rows):\n",
        "       1097    403     |  73.13\n",
        "        607   1117     |  64.79\n",
        "    --------------\n",
        "      64.38  73.49\n",
        "\n",
        "             precision    recall  f1-score   support\n",
        "\n",
        "          0       0.64      0.73      0.68      1500\n",
        "          1       0.73      0.65      0.69      1724\n",
        "\n",
        "avg / total       0.69      0.69      0.69      3224\n",
        "\n"
       ]
      }
     ],
     "prompt_number": 20
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "figure,(plt1,plt2) = pyplot.subplots(1,2)\n",
      "figure.set_size_inches(10,5)\n",
      "figure.tight_layout()\n",
      "\n",
      "createROCPlot(plt1,test_ys,morgan_probabilities,'Morgan FP')\n",
      "createROCPlot(plt2,test_ys,rd_probabilities,'RDKit FP')\n",
      "\n",
      "figure,(plt1,plt2) = pyplot.subplots(1,2)\n",
      "figure.set_size_inches(10,5)\n",
      "figure.tight_layout()\n",
      "\n",
      "createPropAccPlot(plt1,test_ys,morgan_probabilities,'Morgan FP')\n",
      "createPropAccPlot(plt2,test_ys,rd_probabilities,'RDKit FP')\n",
      "\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Area under the ROC curve : 0.789042\n",
        "Area under the ROC curve : 0.733286\n"
       ]
      },
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAtUAAAF5CAYAAABQhdZjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nsakmIiIiIqohNtVERERERDX0/1Szr5TnAgjCAAAAAElFTkSuQmCC\n"
      },
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAtcAAAF7CAYAAAAZuKdVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nERERkUo4uSYiIiIiUgkn10REREREKvkfRclGtar+Nc0AAAAASUVORK5CYII=\n"
      }
     ],
     "prompt_number": 21
    },
    {
     "cell_type": "heading",
     "level": 4,
     "metadata": {},
     "source": [
      "Obtain bit probabilities for later analysis"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": true,
     "input": [
      "bit_probabilities = morgan_bnb.feature_log_prob_\n",
      "position = []\n",
      "enrichment = []\n",
      "for d in zip(range(len(bit_probabilities[1])),[(w[0]-w[1]) for w in zip(bit_probabilities[0],bit_probabilities[1])]):\n",
      "    if abs(d[1]) >= 1: #one log-level probablity difference between classes\n",
      "        position.append(d[0])\n",
      "        enrichment.append(d[1])\n",
      " "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 22
    },
    {
     "cell_type": "heading",
     "level": 4,
     "metadata": {},
     "source": [
      "Binary classification is obtained by mapping selecting the most likely class "
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#as implemented in sklearn\n",
      "check_predictions = [1 if t[1] >= t[0] else 0 for t in morgan_bnb.predict_proba(test_morgan_fps)]\n",
      "print confusion_matrix_summary(test_ys,check_predictions)\n",
      "\n",
      "#equivalent for binary classification\n",
      "check_predictions = [1 if p >= 0.5 else 0 for p in [t[1] for t in morgan_bnb.predict_proba(test_morgan_fps)]]\n",
      "print confusion_matrix_summary(test_ys,check_predictions)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "\n",
        "\tResults Table (experiment in rows):\n",
        "       1132    368     |  75.47\n",
        "        530   1194     |  69.26\n",
        "    --------------\n",
        "      68.11  76.44\n",
        "\n",
        "\n"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "\tResults Table (experiment in rows):\n",
        "       1132    368     |  75.47\n",
        "        530   1194     |  69.26\n",
        "    --------------\n",
        "      68.11  76.44\n",
        "\n"
       ]
      }
     ],
     "prompt_number": 23
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "Random Forests"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": true,
     "input": [
      "from sklearn.cross_validation import KFold\n",
      "from sklearn.grid_search import GridSearchCV\n",
      "from sklearn.ensemble import RandomForestClassifier\n",
      "#params = {'max_depth':[2,5,10,20],'min_samples_split':[2,8,32,128],'min_samples_leaf':[1,2,5,10],'compute_importances':[True],'n_estimators':[200]}\n",
      "#grid search is very timeconsuming, ~3hrs\n",
      "params = {'compute_importances': [True], 'min_samples_split': [8], 'max_depth': [20], 'min_samples_leaf': [1],'n_estimators':[200]}\n",
      "cv = KFold(n=len(train),k=10,indices=False,shuffle=True)\n",
      "gs = GridSearchCV(RandomForestClassifier(), params, cv=cv,verbose=1,refit=True)\n",
      "gs.fit(train_morgan_fps, train_ys)\n",
      "\n",
      "#print 'Best score: %0.2f'%gs.best_score_\n",
      "#print 'Training set performance using best parameters (%s)'%gs.best_params_\n",
      "best_morgan_treemodel = gs.best_estimator_\n",
      "#training set evaluation\n",
      "best_morgan_tree_prediction = best_morgan_treemodel.predict(test_morgan_fps)\n",
      "print 'Morgan Fingerprints'\n",
      "print confusion_matrix_summary(test_ys,best_morgan_tree_prediction)\n",
      "print classification_report(test_ys,best_morgan_tree_prediction)\n",
      "\n",
      "print '------------------------------------------------------------'\n",
      "gs.fit(train_rd_fps, train_ys)\n",
      "#print 'Best score: %0.2f'%gs.best_score_\n",
      "#print 'Training set performance using best parameters (%s)'%gs.best_params_\n",
      "best_rd_treemodel = gs.best_estimator_\n",
      "#training set evaluation\n",
      "best_rd_tree_prediction = best_rd_treemodel.predict(test_rd_fps)\n",
      "print 'RDKit Fingerprints'\n",
      "print confusion_matrix_summary(test_ys,best_rd_tree_prediction)\n",
      "print classification_report(test_ys,best_rd_tree_prediction)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Morgan Fingerprints\n",
        "\n",
        "\n",
        "\tResults Table (experiment in rows):\n",
        "       1172    328     |  78.13\n",
        "        449   1275     |  73.96\n",
        "    --------------\n",
        "      72.30  79.54\n",
        "\n",
        "             precision    recall  f1-score   support\n",
        "\n",
        "          0       0.72      0.78      0.75      1500\n",
        "          1       0.80      0.74      0.77      1724\n",
        "\n",
        "avg / total       0.76      0.76      0.76      3224\n",
        "\n",
        "------------------------------------------------------------\n",
        "RDKit Fingerprints"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "\n",
        "\n",
        "\tResults Table (experiment in rows):\n",
        "       1134    366     |  75.60\n",
        "        347   1377     |  79.87\n",
        "    --------------\n",
        "      76.57  79.00\n",
        "\n",
        "             precision    recall  f1-score   support\n",
        "\n",
        "          0       0.77      0.76      0.76      1500\n",
        "          1       0.79      0.80      0.79      1724\n",
        "\n",
        "avg / total       0.78      0.78      0.78      3224\n",
        "\n"
       ]
      }
     ],
     "prompt_number": 24
    },
    {
     "cell_type": "heading",
     "level": 4,
     "metadata": {},
     "source": [
      "Probabilities can be obtained as proportion of the trees that predict the respective class"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "morgan_probabilities = [p[1] for p in best_morgan_treemodel.predict_proba(test_morgan_fps)]\n",
      "rd_probabilities = [p[1] for p in best_rd_treemodel.predict_proba(test_rd_fps)]\n",
      "figure,(plt1,plt2) = pyplot.subplots(1,2)\n",
      "figure.set_size_inches(10,5)\n",
      "figure.tight_layout()\n",
      "\n",
      "createROCPlot(plt1,test_ys,morgan_probabilities,'Morgan FP')\n",
      "createROCPlot(plt2,test_ys,rd_probabilities,'RDKit FP')\n",
      "\n",
      "figure,(plt1,plt2) = pyplot.subplots(1,2)\n",
      "figure.set_size_inches(10,5)\n",
      "figure.tight_layout()\n",
      "\n",
      "createPropAccPlot(plt1,test_ys,morgan_probabilities,'Morgan FP')\n",
      "createPropAccPlot(plt2,test_ys,rd_probabilities,'RDKit FP')\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Area under the ROC curve : 0.838236\n",
        "Area under the ROC curve : 0.854807\n"
       ]
      },
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAtUAAAF5CAYAAABQhdZjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nqomIiIiImuj/AUNWrgLxi4amAAAAAElFTkSuQmCC\n"
      },
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAtcAAAF7CAYAAAAZuKdVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n"
      }
     ],
     "prompt_number": 25
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Substructure analysis"
     ]
    },
    {
     "cell_type": "heading",
     "level": 4,
     "metadata": {},
     "source": [
      "Idea: Compute the probability of a bit for each class, identify bits enriched in one class and report substructures associated with these bits. Then identify substructures that occur predominantly in molecules belonging to one class"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "images = []\n",
      "smiles = {}\n",
      "#find all substructures associated with enriched bits\n",
      "for bit in zip(position,enrichment):\n",
      "    i = 0\n",
      "    class0_prob = bit_probabilities[0][bit[0]]\n",
      "    class1_prob = bit_probabilities[1][bit[0]]\n",
      "    for info in train_info:\n",
      "        if info.has_key(bit[0]):\n",
      "            try:\n",
      "                m = data[train[i]][0]\n",
      "                atomId,radius = info[bit[0]][0]\n",
      "                env=Chem.FindAtomEnvironmentOfRadiusN(m,radius,atomId)\n",
      "                amap={}            \n",
      "                #submol = Chem.PathToSubmol(m,env,atomMap=amap)\n",
      "                ats = set([atomId])\n",
      "                for bidx in env:\n",
      "                    bond = m.GetBondWithIdx(bidx)\n",
      "                    ats.add(bond.GetBeginAtomIdx())\n",
      "                    ats.add(bond.GetEndAtomIdx())\n",
      "                smi = Chem.MolFragmentToSmiles(m,atomsToUse=list(ats),bondsToUse=env,rootedAtAtom=atomId)\n",
      "                #smi = Chem.FastFindRings(smi)\n",
      "                \n",
      "                if smiles.has_key(smi):\n",
      "                    num_molecules = smiles[smi][0]\n",
      "                    num_positives = smiles[smi][1]\n",
      "                    if data[train[i]][1] == 1: num_positives+=1\n",
      "                    smiles[smi] = (num_molecules+1,num_positives,bit[0],class0_prob,class1_prob)\n",
      "                else:\n",
      "                    num_positives = 0\n",
      "                    if data[train[i]][1] == 1: num_positives=1\n",
      "                    smiles[smi] = (1,num_positives,bit[0],class0_prob,class1_prob)\n",
      "                \n",
      "            except:\n",
      "                continue\n",
      "        i+=1\n",
      "\n",
      "        results=[]\n",
      "for smi in smiles.keys():\n",
      "    ratio = float(smiles[smi][1])/smiles[smi][0]\n",
      "    if smiles[smi][0] > 5 and (ratio < 0.2 or ratio > 0.8): #show only substructures occuring in at least 5 molecules, with either less than 20% or more than 80% active proportion\n",
      "        results.append((smiles[smi][2],smiles[smi][0],smiles[smi][1],(smiles[smi][3]),(smiles[smi][4]),smi))\n",
      "results.sort(key=lambda x: x[0])\n",
      "for line in results:        \n",
      "    print 'bit %i: \\t%i (%i) molecules, in-class log_probabilities %0.3f %0.3f \\t SMILES: %s'%line#(smiles[smi][2],smiles[smi][0],smiles[smi][1],(smiles[smi][3]),(smiles[smi][4]),smi)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "bit 4: \t7 (0) molecules, in-class log_probabilities -7.357 -9.039 \t SMILES: S(=O)(=O)(Nc)c(c)c\n",
        "bit 48: \t13 (13) molecules, in-class log_probabilities -9.266 -7.696 \t SMILES: C1(c(c)c)OC1C\n",
        "bit 105: \t6 (1) molecules, in-class log_probabilities -7.819 -8.885 \t SMILES: c(nn)c(c)Cl\n",
        "bit 121: \t8 (7) molecules, in-class log_probabilities -8.861 -7.835 \t SMILES: c(c)(-c)o\n",
        "bit 121: \t6 (5) molecules, in-class log_probabilities -8.861 -7.835 \t SMILES: c(-c)(c)o\n",
        "bit 131: \t11 (10) molecules, in-class log_probabilities -9.266 -8.192 \t SMILES: c(c)(c)s\n",
        "bit 164: \t16 (15) molecules, in-class log_probabilities -7.069 -6.011 \t SMILES: c(c)([N+])s\n",
        "bit 192: \t8 (0) molecules, in-class log_probabilities -6.478 -7.499 \t SMILES: C(Oc)C(N)=O\n",
        "bit 199: \t32 (29) molecules, in-class log_probabilities -8.088 -6.206 \t SMILES: c(c(c)[N+])c(c)[N+]\n",
        "bit 227: \t6 (1) molecules, in-class log_probabilities -6.153 -7.194 \t SMILES: C(c)=C\n",
        "bit 254: \t7 (1) molecules, in-class log_probabilities -8.014 -9.039 \t SMILES: C(C)(C)CO\n",
        "bit 255: \t36 (36) molecules, in-class log_probabilities -8.168 -7.003 \t SMILES: [N+](=O)([O-])c(c)o\n",
        "bit 256: \t16 (16) molecules, in-class log_probabilities -8.707 -7.499 \t SMILES: c(cc)(C=C)oc\n",
        "bit 270: \t14 (12) molecules, in-class log_probabilities -8.255 -7.168 \t SMILES: c(cc)(cc)C=N\n",
        "bit 270: \t23 (22) molecules, in-class log_probabilities -8.255 -7.168 \t SMILES: c(cc)c(C)o\n",
        "bit 298: \t11 (11) molecules, in-class log_probabilities -9.043 -7.535 \t SMILES: c(cc)(c(c)C)C(c)=O\n",
        "bit 298: \t12 (11) molecules, in-class log_probabilities -9.043 -7.535 \t SMILES: c(cc)(c(C)c)C(c)=O\n",
        "bit 323: \t11 (9) molecules, in-class log_probabilities -8.255 -7.142 \t SMILES: c(c)(-c)n\n",
        "bit 358: \t13 (13) molecules, in-class log_probabilities -9.266 -7.835 \t SMILES: O1C(c)C1c\n",
        "bit 374: \t6 (5) molecules, in-class log_probabilities -9.043 -7.887 \t SMILES: C(OC)(Oc)C(c)C\n",
        "bit 394: \t16 (3) molecules, in-class log_probabilities -7.819 -9.039 \t SMILES: CS\n",
        "bit 460: \t6 (6) molecules, in-class log_probabilities -9.043 -7.787 \t SMILES: c(NC)(c(c)c)c(c)c\n",
        "bit 460: \t8 (8) molecules, in-class log_probabilities -9.043 -7.787 \t SMILES: c(cc)(c(-c)c)C(C)N\n",
        "bit 505: \t6 (5) molecules, in-class log_probabilities -9.266 -8.266 \t SMILES: C(CC)(OC)c(c)c\n",
        "bit 584: \t19 (18) molecules, in-class log_probabilities -8.861 -7.787 \t SMILES: C(=Cc)c(c)c\n",
        "bit 587: \t16 (16) molecules, in-class log_probabilities -8.573 -7.094 \t SMILES: c(cc)(cc)C(N)=O\n",
        "bit 604: \t7 (0) molecules, in-class log_probabilities -8.255 -10.138 \t SMILES: c(cc)(nc)N(C)C\n",
        "bit 628: \t9 (1) molecules, in-class log_probabilities -7.657 -9.039 \t SMILES: C[N+]\n",
        "bit 647: \t56 (47) molecules, in-class log_probabilities -8.014 -6.880 \t SMILES: c(N)(cc)cc\n",
        "bit 669: \t10 (0) molecules, in-class log_probabilities -8.168 -10.138 \t SMILES: C(OC)(OC)C(C)O\n",
        "bit 684: \t12 (1) molecules, in-class log_probabilities -7.708 -9.445 \t SMILES: C(O)(=O)CO\n",
        "bit 711: \t10 (9) molecules, in-class log_probabilities -8.861 -7.573 \t SMILES: c(cc)(c(c)c)n(C)s\n",
        "bit 711: \t11 (9) molecules, in-class log_probabilities -8.861 -7.573 \t SMILES: O(CO)c(c)c\n",
        "bit 715: \t538 (437) molecules, in-class log_probabilities -5.733 -4.722 \t SMILES: [O-]\n",
        "bit 716: \t211 (177) molecules, in-class log_probabilities -6.964 -5.605 \t SMILES: c(c)(c)[N+]\n",
        "bit 724: \t6 (1) molecules, in-class log_probabilities -7.608 -8.634 \t SMILES: C(c)(c)=N\n",
        "bit 740: \t7 (7) molecules, in-class log_probabilities -8.861 -7.499 \t SMILES: [N-]=[N+]\n",
        "bit 753: \t510 (424) molecules, in-class log_probabilities -6.099 -4.767 \t SMILES: O=[N+]\n",
        "bit 785: \t16 (13) molecules, in-class log_probabilities -9.266 -8.192 \t SMILES: c(cc)(cc)-c(c)c\n",
        "bit 811: \t10 (10) molecules, in-class log_probabilities -8.707 -7.696 \t SMILES: O(C(C)=O)N(C)O\n",
        "bit 815: \t24 (1) molecules, in-class log_probabilities -8.168 -6.474 \t SMILES: S(c)(N)(=O)=O\n",
        "bit 815: \t99 (92) molecules, in-class log_probabilities -8.168 -6.474 \t SMILES: N(N)=O\n",
        "bit 824: \t14 (12) molecules, in-class log_probabilities -8.707 -9.733 \t SMILES: N(=O)N(C)C\n",
        "bit 838: \t521 (430) molecules, in-class log_probabilities -6.120 -4.763 \t SMILES: [O-][N+]\n",
        "bit 842: \t9 (9) molecules, in-class log_probabilities -9.554 -7.887 \t SMILES: c(cc)(c(c)n)c(=N)s\n",
        "bit 860: \t47 (43) molecules, in-class log_probabilities -8.861 -7.696 \t SMILES: c(c(c)c)c(c)[N+]\n",
        "bit 870: \t16 (16) molecules, in-class log_probabilities -8.573 -7.535 \t SMILES: c(C)(c(c)c)c(c)c\n",
        "bit 870: \t6 (5) molecules, in-class log_probabilities -8.573 -7.535 \t SMILES: c(cc)(c(c)c)[N+](=O)[O-]\n",
        "bit 909: \t29 (25) molecules, in-class log_probabilities -7.475 -9.039 \t SMILES: N(C)(C)N\n",
        "bit 930: \t6 (0) molecules, in-class log_probabilities -8.573 -10.138 \t SMILES: c(Cl)(nc)nc\n",
        "bit 957: \t16 (15) molecules, in-class log_probabilities -9.266 -7.653 \t SMILES: [N+](=O)([O-])c(c)s\n",
        "bit 967: \t9 (9) molecules, in-class log_probabilities -8.014 -9.222 \t SMILES: c(cc)(cs)C(N)=O\n",
        "bit 979: \t126 (115) molecules, in-class log_probabilities -7.657 -5.995 \t SMILES: O=N\n",
        "bit 979: \t33 (3) molecules, in-class log_probabilities -7.657 -5.995 \t SMILES: N(C)C\n",
        "bit 991: \t8 (0) molecules, in-class log_probabilities -7.945 -9.733 \t SMILES: c(cc)(OC)c(c)Cl\n",
        "bit 994: \t7 (7) molecules, in-class log_probabilities -9.266 -7.573 \t SMILES: n(sc)(c(c)c)C(c)=O\n",
        "bit 994: \t12 (12) molecules, in-class log_probabilities -9.266 -7.573 \t SMILES: C(ON)c(c)c\n",
        "bit 1011: \t39 (7) molecules, in-class log_probabilities -7.126 -8.433 \t SMILES: c(c)(c)Cl\n",
        "bit 1050: \t6 (0) molecules, in-class log_probabilities -8.014 -10.138 \t SMILES: FC\n",
        "bit 1050: \t6 (0) molecules, in-class log_probabilities -8.014 -10.138 \t SMILES: C(O)(CC)CC\n",
        "bit 1056: \t32 (32) molecules, in-class log_probabilities -8.350 -7.094 \t SMILES: c(cc)c([N+])o\n",
        "bit 1064: \t14 (1) molecules, in-class log_probabilities -8.014 -9.445 \t SMILES: N(=C)O\n",
        "bit 1069: \t7 (0) molecules, in-class log_probabilities -8.255 -10.138 \t SMILES: C(CO)OC\n",
        "bit 1091: \t9 (0) molecules, in-class log_probabilities -8.455 -9.733 \t SMILES: O(c(c)c)P(O)(O)=O\n",
        "bit 1098: \t20 (17) molecules, in-class log_probabilities -9.266 -7.696 \t SMILES: N(CC)c(c)c\n",
        "bit 1103: \t6 (6) molecules, in-class log_probabilities -9.266 -8.059 \t SMILES: n(C)(c)s\n",
        "bit 1131: \t6 (1) molecules, in-class log_probabilities -9.960 -8.529 \t SMILES: c(cc)(CC)c(C)c\n",
        "bit 1155: \t19 (2) molecules, in-class log_probabilities -7.708 -9.733 \t SMILES: c(cc)cn\n",
        "bit 1166: \t7 (6) molecules, in-class log_probabilities -9.266 -8.192 \t SMILES: c(N)(c(c)c)c(c)c\n",
        "bit 1166: \t6 (5) molecules, in-class log_probabilities -9.266 -8.192 \t SMILES: c(C)(cc)c(c)n\n",
        "bit 1188: \t7 (6) molecules, in-class log_probabilities -9.960 -8.885 \t SMILES: BrC\n",
        "bit 1195: \t477 (404) molecules, in-class log_probabilities -6.198 -4.803 \t SMILES: [N+](c)(=O)[O-]\n",
        "bit 1216: \t34 (34) molecules, in-class log_probabilities -9.960 -7.142 \t SMILES: c(cc)(oc)[N+](=O)[O-]\n",
        "bit 1221: \t22 (21) molecules, in-class log_probabilities -9.043 -7.653 \t SMILES: C(c)(N)=O\n",
        "bit 1256: \t24 (3) molecules, in-class log_probabilities -7.562 -9.222 \t SMILES: C(CC)C(O)=O\n",
        "bit 1256: \t10 (0) molecules, in-class log_probabilities -7.562 -9.222 \t SMILES: C(C)=O\n",
        "bit 1264: \t15 (15) molecules, in-class log_probabilities -8.350 -7.220 \t SMILES: c(sc)c(-c)n\n",
        "bit 1270: \t16 (2) molecules, in-class log_probabilities -7.880 -9.445 \t SMILES: N(c)S\n",
        "bit 1287: \t50 (7) molecules, in-class log_probabilities -6.989 -8.529 \t SMILES: C(O)(=O)CC\n",
        "bit 1299: \t28 (23) molecules, in-class log_probabilities -8.707 -7.612 \t SMILES: c(cc)([nH]c)c(c)c\n",
        "bit 1300: \t8 (7) molecules, in-class log_probabilities -8.088 -7.047 \t SMILES: C(Cc)C(c)=O\n",
        "bit 1300: \t37 (34) molecules, in-class log_probabilities -8.088 -7.047 \t SMILES: n(c(c)c)c(c)c\n",
        "bit 1305: \t10 (9) molecules, in-class log_probabilities -9.554 -8.266 \t SMILES: c(c)(=N)s\n",
        "bit 1326: \t7 (0) molecules, in-class log_probabilities -8.088 -10.138 \t SMILES: C(CC)(C(C)C)C(C)C\n",
        "bit 1339: \t13 (0) molecules, in-class log_probabilities -7.475 -9.445 \t SMILES: C(C)(C)(C)c\n",
        "bit 1348: \t6 (5) molecules, in-class log_probabilities -8.350 -6.737 \t SMILES: c(nc)(c(c)n)c(c)n\n",
        "bit 1358: \t6 (1) molecules, in-class log_probabilities -8.455 -9.733 \t SMILES: C(C)Nc\n",
        "bit 1386: \t6 (0) molecules, in-class log_probabilities -6.610 -8.059 \t SMILES: C(CC)C(C)=C\n",
        "bit 1415: \t13 (1) molecules, in-class log_probabilities -7.562 -8.634 \t SMILES: C(C)(C)(C)c(c)c\n",
        "bit 1415: \t474 (404) molecules, in-class log_probabilities -7.562 -8.634 \t SMILES: c(cc)(cc)c(c)c\n",
        "bit 1421: \t7 (1) molecules, in-class log_probabilities -8.455 -10.831 \t SMILES: O(CC)P(O)(O)=O\n",
        "bit 1434: \t9 (1) molecules, in-class log_probabilities -8.707 -10.831 \t SMILES: c(cc)(cc)C(C)O\n",
        "bit 1444: \t261 (38) molecules, in-class log_probabilities -6.494 -7.653 \t SMILES: C(CC)CC\n",
        "bit 1482: \t9 (0) molecules, in-class log_probabilities -7.762 -8.885 \t SMILES: N(CC)(CC)CC\n",
        "bit 1489: \t17 (1) molecules, in-class log_probabilities -9.043 -10.831 \t SMILES: c(cc)(cc)S(N)(=O)=O\n",
        "bit 1519: \t18 (1) molecules, in-class log_probabilities -8.861 -10.138 \t SMILES: C(CC)C(C)(C)C\n",
        "bit 1522: \t7 (1) molecules, in-class log_probabilities -7.434 -8.752 \t SMILES: c(cc)(cc)C(F)(F)F\n",
        "bit 1523: \t47 (43) molecules, in-class log_probabilities -8.455 -6.981 \t SMILES: C1OC1C\n",
        "bit 1538: \t7 (1) molecules, in-class log_probabilities -8.455 -10.138 \t SMILES: C(CC)(C(C)(C)C)C(C)(C)C\n",
        "bit 1550: \t38 (7) molecules, in-class log_probabilities -7.880 -9.445 \t SMILES: C(C)(C)=O\n",
        "bit 1554: \t32 (6) molecules, in-class log_probabilities -7.517 -8.752 \t SMILES: c(Cl)(cc)c(c)Cl\n",
        "bit 1554: \t6 (0) molecules, in-class log_probabilities -7.517 -8.752 \t SMILES: C(=O)CC\n",
        "bit 1566: \t26 (25) molecules, in-class log_probabilities -9.960 -8.529 \t SMILES: c(cc)(c(c)c)c(c)n\n",
        "bit 1571: \t6 (0) molecules, in-class log_probabilities -8.255 -10.138 \t SMILES: S(c)(c)(=O)=O\n",
        "bit 1572: \t9 (8) molecules, in-class log_probabilities -9.266 -8.266 \t SMILES: n(cn)c(c)N\n",
        "bit 1574: \t264 (235) molecules, in-class log_probabilities -7.657 -8.885 \t SMILES: c(cc)(c(c)c)c(c)c\n",
        "bit 1574: \t6 (0) molecules, in-class log_probabilities -7.657 -8.885 \t SMILES: N(CC)C(C)C\n",
        "bit 1579: \t11 (0) molecules, in-class log_probabilities -9.266 -10.831 \t SMILES: C(C=C)C(C)C\n",
        "bit 1589: \t7 (1) molecules, in-class log_probabilities -7.880 -9.445 \t SMILES: C(=O)(CC)NC\n",
        "bit 1592: \t7 (0) molecules, in-class log_probabilities -7.285 -10.138 \t SMILES: c(Cl)(cc)c(c)S\n",
        "bit 1611: \t6 (1) molecules, in-class log_probabilities -8.707 -10.831 \t SMILES: O(C(C)=O)C(C)(C)C\n",
        "bit 1615: \t11 (10) molecules, in-class log_probabilities -9.960 -7.887 \t SMILES: c(cc)(c(c)c)c(c)[nH]\n",
        "bit 1624: \t20 (3) molecules, in-class log_probabilities -7.285 -8.529 \t SMILES: C(O)(=O)C(C)N\n",
        "bit 1624: \t38 (6) molecules, in-class log_probabilities -7.285 -8.529 \t SMILES: C(C)(N)=O\n",
        "bit 1660: \t7 (1) molecules, in-class log_probabilities -8.707 -10.138 \t SMILES: c(cc)(CC)c(c)C\n",
        "bit 1672: \t7 (0) molecules, in-class log_probabilities -8.168 -10.138 \t SMILES: C(OC)(C(C)O)C(O)=O\n",
        "bit 1673: \t46 (5) molecules, in-class log_probabilities -7.156 -8.885 \t SMILES: C(=O)(OC)c(c)c\n",
        "bit 1674: \t13 (13) molecules, in-class log_probabilities -9.960 -8.059 \t SMILES: C1(c(c)c)OC1c\n",
        "bit 1694: \t6 (6) molecules, in-class log_probabilities -9.960 -8.059 \t SMILES: C1(c(c)c)NC1c\n",
        "bit 1704: \t23 (4) molecules, in-class log_probabilities -9.266 -8.192 \t SMILES: C(C)(C)(C)N\n",
        "bit 1733: \t10 (9) molecules, in-class log_probabilities -6.846 -5.558 \t SMILES: c(cc)(nc)c(c)n\n",
        "bit 1734: \t42 (6) molecules, in-class log_probabilities -7.126 -8.634 \t SMILES: C(CC)OC\n",
        "bit 1737: \t7 (0) molecules, in-class log_probabilities -6.309 -7.612 \t SMILES: c(c(C)c)c(c)O\n",
        "bit 1740: \t26 (1) molecules, in-class log_probabilities -9.043 -10.831 \t SMILES: C(=O)(CC)OC\n",
        "bit 1768: \t39 (5) molecules, in-class log_probabilities -7.708 -8.752 \t SMILES: C(C)(C)=C\n",
        "bit 1768: \t8 (0) molecules, in-class log_probabilities -7.708 -8.752 \t SMILES: c(cc)(c(c)O)C(C)(C)C\n",
        "bit 1795: \t11 (11) molecules, in-class log_probabilities -9.266 -7.887 \t SMILES: C(=O)(Nc)c(c)c\n",
        "bit 1796: \t9 (0) molecules, in-class log_probabilities -7.880 -8.885 \t SMILES: C(=O)(CC)Nc\n",
        "bit 1806: \t6 (6) molecules, in-class log_probabilities -9.960 -8.346 \t SMILES: C(=Nn)c(c)c\n",
        "bit 1809: \t431 (366) molecules, in-class log_probabilities -7.097 -5.569 \t SMILES: [N+](=O)([O-])c(c)c\n",
        "bit 1814: \t297 (259) molecules, in-class log_probabilities -7.434 -5.719 \t SMILES: c(cc)c(c)[N+]\n",
        "bit 1822: \t7 (0) molecules, in-class log_probabilities -8.707 -7.573 \t SMILES: c(cc)(cc)CC\n",
        "bit 1822: \t19 (19) molecules, in-class log_probabilities -8.707 -7.573 \t SMILES: C(CC)N(C)N\n",
        "bit 1822: \t6 (6) molecules, in-class log_probabilities -8.707 -7.573 \t SMILES: c(c(c)N)c(c)[N+]\n",
        "bit 1822: \t26 (3) molecules, in-class log_probabilities -8.707 -7.573 \t SMILES: O(CC)C(c)=O\n",
        "bit 1829: \t187 (34) molecules, in-class log_probabilities -8.168 -9.445 \t SMILES: C(C)(O)=O\n",
        "bit 1829: \t7 (0) molecules, in-class log_probabilities -8.168 -9.445 \t SMILES: C(C)(C)(F)F\n",
        "bit 1884: \t6 (6) molecules, in-class log_probabilities -7.819 -9.222 \t SMILES: c(cc)(c(c)c)C(C)O\n",
        "bit 1893: \t7 (7) molecules, in-class log_probabilities -9.554 -8.123 \t SMILES: C(C)(c)N\n",
        "bit 1893: \t8 (7) molecules, in-class log_probabilities -9.554 -8.123 \t SMILES: C(c)(C)N\n",
        "bit 1894: \t17 (1) molecules, in-class log_probabilities -9.554 -10.831 \t SMILES: C(CC)=C(C)C\n",
        "bit 1899: \t7 (6) molecules, in-class log_probabilities -9.266 -7.887 \t SMILES: C(O)(C(C)O)C(c)O\n",
        "bit 1899: \t21 (3) molecules, in-class log_probabilities -9.266 -7.887 \t SMILES: C(O)(C(C)O)C(C)O\n",
        "bit 1899: \t12 (11) molecules, in-class log_probabilities -9.266 -7.887 \t SMILES: C(O)(C(c)O)C(C)O\n",
        "bit 1912: \t119 (104) molecules, in-class log_probabilities -8.088 -6.627 \t SMILES: c(c(c)c)(c(c)c)c(c)c\n",
        "bit 1914: \t7 (6) molecules, in-class log_probabilities -8.088 -6.880 \t SMILES: C(CCl)N(C)C\n",
        "bit 1914: \t120 (107) molecules, in-class log_probabilities -8.088 -6.880 \t SMILES: c(c(c)c)c(c)c\n",
        "bit 1916: \t7 (0) molecules, in-class log_probabilities -8.350 -10.831 \t SMILES: c(cc)(c(c)N)C(O)=O\n",
        "bit 1918: \t23 (4) molecules, in-class log_probabilities -8.014 -9.445 \t SMILES: c(Cl)(cc)c(c)O\n",
        "bit 1918: \t7 (1) molecules, in-class log_probabilities -8.014 -9.445 \t SMILES: c(c)(n)N\n",
        "bit 1923: \t24 (24) molecules, in-class log_probabilities -7.708 -6.569 \t SMILES: o(c(c)C)c(c)[N+]\n",
        "bit 1925: \t74 (60) molecules, in-class log_probabilities -8.455 -7.094 \t SMILES: C(=O)(c(c)c)c(c)c\n",
        "bit 1934: \t8 (8) molecules, in-class log_probabilities -8.861 -7.835 \t SMILES: c(-c)(n)s\n",
        "bit 1936: \t6 (5) molecules, in-class log_probabilities -8.861 -7.740 \t SMILES: c(c(c)C)c(c)c\n",
        "bit 1936: \t7 (7) molecules, in-class log_probabilities -8.861 -7.740 \t SMILES: c(c(c)C)c(c)N\n",
        "bit 1940: \t199 (169) molecules, in-class log_probabilities -7.562 -6.196 \t SMILES: c(cc)(cc)[N+](=O)[O-]\n",
        "bit 1940: \t16 (1) molecules, in-class log_probabilities -7.562 -6.196 \t SMILES: C(NC)C(C)O\n",
        "bit 1955: \t6 (0) molecules, in-class log_probabilities -8.088 -9.733 \t SMILES: C(CC)C(N)=O\n",
        "bit 1963: \t8 (0) molecules, in-class log_probabilities -6.088 -4.754 \t SMILES: O(C(C)C)C(C)O\n",
        "bit 1963: \t524 (432) molecules, in-class log_probabilities -6.088 -4.754 \t SMILES: [N+]\n",
        "bit 1984: \t304 (255) molecules, in-class log_probabilities -6.664 -5.204 \t SMILES: c(c)(c)c\n",
        "bit 1984: \t18 (16) molecules, in-class log_probabilities -6.664 -5.204 \t SMILES: O(C)N\n",
        "bit 1989: \t29 (3) molecules, in-class log_probabilities -7.657 -9.039 \t SMILES: N(C)(C)C\n",
        "bit 2013: \t35 (34) molecules, in-class log_probabilities -9.554 -8.433 \t SMILES: c(cc)(c(c)c)c(c)[N+]\n",
        "bit 2016: \t10 (0) molecules, in-class log_probabilities -8.088 -9.445 \t SMILES: N(C(C)C)C(C)=O\n",
        "bit 2016: \t16 (13) molecules, in-class log_probabilities -8.088 -9.445 \t SMILES: C(OC)C(C)O\n",
        "bit 2030: \t14 (12) molecules, in-class log_probabilities -8.707 -7.535 \t SMILES: c(cc)(c(c)c)c(c)C\n",
        "bit 2030: \t15 (15) molecules, in-class log_probabilities -8.707 -7.535 \t SMILES: c(cc)(c(c)C)c(c)c\n",
        "bit 2030: \t7 (1) molecules, in-class log_probabilities -8.707 -7.535 \t SMILES: C(C)(C)NC\n",
        "bit 2031: \t7 (1) molecules, in-class log_probabilities -8.255 -9.733 \t SMILES: C(CN)N(c)C\n",
        "bit 2036: \t9 (0) molecules, in-class log_probabilities -8.255 -9.445 \t SMILES: C(CC)C(C)N\n",
        "bit 2038: \t164 (30) molecules, in-class log_probabilities -6.964 -7.998 \t SMILES: C(C)(C)C\n"
       ]
      }
     ],
     "prompt_number": 26
    }
   ],
   "metadata": {}
  }
 ]
}