notebooks/Substructure Fingerprints.ipynb

{
 "metadata": {
  "name": ""
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Fingerprint-based substructure screening 1\n",
      "\n",
      "Some form of fingerprint is often used to make substructure searching more efficient. The idea is to use a fingerprinting algorithm with the property:\n",
      "```\n",
      "FP(query) & FP(mol) = FP(query)\n",
      "```\n",
      "if query is a substructure of mol. In words: if `query` is a substructure of `mol` then every bit set in the fingerprint of `query` should also be set in `mol`.\n",
      "\n",
      "A bunch of different approaches to this have been developed, I'm not going to review them all here. :-)\n",
      "\n",
      "Andrew Dalke has done some writing on the topic: http://dalkescientific.com/writings/diary/archive/2012/06/11/optimizing_substructure_keys.html\n",
      "\n",
      "One of the best-known approaches is the Daylight algorithm:\n",
      "http://www.daylight.com/dayhtml/doc/theory/theory.finger.html\n",
      "\n",
      "As Andrew mentions in the post I link to above, one of the big problems here is the lack of a reasonable query dataset for benchmarking. The approach I've taken is to use collections of small molecules from ZINC as well as some fragments of PubChem molecules. The database I query is a set of ChEMBL molecules from an earlier post.\n",
      "\n",
      "Here's the atom-number count distributions of the queries and molecules:\n",
      "\n",
      "<img src=\"data:image/png;base64,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\"/>\n",
      "\n",
      "It's not perfect, but it's a start.\n",
      "\n",
      "## TL;DR summary\n",
      "\n",
      "Using the RDKit pattern fingerprinter (the default used in the postgresql cartridge, more on that in another post), a screenout accuracy of around 60% is achievable with these three query sets. This means that 60% of molecules passing the fingerprint screen actually have a substructure match.\n",
      "\n",
      "To put that in perspective: with the leads query set and the pattern fingerprint, 4650 substructure matches are found in a total of 7601 searches, that's 61% accuracy for the fingerprint. If the fingerprint had not been used to pre-screen, 25000000 (500*50000) searches would have needed to be done, so we've reduced the number of substructure searches by 99.97%. By this logic, even a comparatively poor fingerprint with an accuracy of 5% would help enormously: reducing the number of searches by 99.7%. This will become important when actually using the fingerprints in a database index, but that's a different post.\n",
      "\n",
      "On the technical side: there's some info below about using IPython's ability to run code in parallel.\n"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from rdkit import Chem\n",
      "from rdkit.Chem import rdMolDescriptors\n",
      "from rdkit.Avalon import pyAvalonTools\n",
      "from rdkit.Chem import Draw\n",
      "from rdkit.Chem.Draw import IPythonConsole\n",
      "from rdkit import rdBase\n",
      "from rdkit import DataStructs\n",
      "import cPickle,random,gzip,time\n",
      "from __future__ import print_function\n",
      "print(rdBase.rdkitVersion)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "2014.03.1pre\n"
       ]
      }
     ],
     "prompt_number": 1
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Read in the data\n",
      "\n",
      "## Database molecules\n",
      "\n",
      "Here we will use the 50K molecules that make up the set of 25K reference pairs generated in an earlier post: http://rdkit.blogspot.ch/2013/10/building-similarity-comparison-set-goal.html\n",
      "\n",
      "As a quick reminder: these are pairs of molecules taken from ChEMBL with MW<600 and a count-based MFP0 similarity of at least 0.7 to each other."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "ind = [x.split() for x in gzip.open('../data/chembl16_25K.pairs.txt.gz')]\n",
      "mols = []\n",
      "for i,row in enumerate(ind):\n",
      "    m1 = Chem.MolFromSmiles(row[1])\n",
      "    mols.append(m1)\n",
      "    m2 = Chem.MolFromSmiles(row[3])\n",
      "    mols.append(m2)\n",
      "    "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 2
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## Query molecules\n",
      "\n",
      "We'll use three sets:\n",
      "\n",
      "   1. Fragments: 500 diverse molecules taken from the ZINC Fragments set\n",
      "   2. Leads: 500 diverse molecules taken from the ZINC Lead-like set\n",
      "   3. Pieces: 823 pieces of molecules obtained by doing a BRICS fragmentation of some molecules from the pubchem screening set.\n",
      "   \n",
      "These sets were discussed in this thread on the mailing list:\n",
      "http://www.mail-archive.com/rdkit-discuss@lists.sourceforge.net/msg02066.html\n",
      "and this presentation:\n",
      "http://www.hinxton.wellcome.ac.uk/advancedcourses/MIOSS%20Greg%20Landrum.pdf"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "frags = [Chem.MolFromSmiles(x.split()[0]) for x in file('../data/zinc.frags.500.q.smi')]\n",
      "leads = [Chem.MolFromSmiles(x.split()[0]) for x in file('../data/zinc.leads.500.q.smi')]\n",
      "pieces = [Chem.MolFromSmiles(x) for x in file('../data/fragqueries.q.txt')]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Look at the size distributions:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "figure(figsize=(16,4))\n",
      "subplot('141')\n",
      "hist([x.GetNumAtoms() for x in pieces],bins=20)\n",
      "title('Pieces')\n",
      "xlabel('NumAtoms')\n",
      "subplot('142')\n",
      "hist([x.GetNumAtoms() for x in frags],bins=20)\n",
      "title('Fragments')\n",
      "xlabel('NumAtoms')\n",
      "subplot('143')\n",
      "hist([x.GetNumAtoms() for x in leads],bins=20)\n",
      "title('Leads')\n",
      "_=xlabel('NumAtoms')\n",
      "subplot('144')\n",
      "hist([x.GetNumAtoms() for x in mols],bins=20)\n",
      "title('Mols')\n",
      "_=xlabel('NumAtoms')\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAA6YAAAEVCAYAAADzQ/gZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nOdV/ACIsw152wS8BAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x64b9c250>"
       ]
      }
     ],
     "prompt_number": 97
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Setting up the parallelization\n",
      "\n",
      "I'm going to be doing a fair amount of embarassingly parallel computation, so I'll take advantage of IPython's support for parallel computing. For this to work, you need to have a cluster of workers set up. This is easy in the Notebook (there's a tab in the dashboard), and it's doable from the command line (http://ipython.org/ipython-doc/stable/parallel/parallel_process.html)\n",
      "\n",
      "It took a bit of time to figure out how to do this, but once over that hurdle, this thing is super easy and useful."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from IPython.parallel import Client\n",
      "rc = Client()\n",
      "dview = rc[:]\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "with dview.sync_imports():\n",
      "    from rdkit import Chem\n",
      "    from rdkit import DataStructs\n",
      "    from rdkit.Avalon import pyAvalonTools"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "importing Chem from rdkit on engine(s)\n",
        "importing DataStructs from rdkit on engine(s)\n",
        "importing pyAvalonTools from rdkit.Avalon on engine(s)\n"
       ]
      }
     ],
     "prompt_number": 5
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Verify that it works:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "t1=time.time()\n",
      "_ = [Chem.RDKFingerprint(m) for m in mols[:10000]]\n",
      "t2=time.time()\n",
      "print('Serial: %.2f'%(t2-t1))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Serial: 15.39\n"
       ]
      }
     ],
     "prompt_number": 28
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "fn = lambda x:Chem.RDKFingerprint(x)\n",
      "t1=time.time()\n",
      "_ = dview.map_sync(fn,mols[:10000])\n",
      "t2=time.time()\n",
      "print('Parallel: %.2f'%(t2-t1))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Parallel: 5.24\n"
       ]
      }
     ],
     "prompt_number": 70
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The above runtimes were collected with a cluster of 4 workers, so it looks like the parallization is working."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# The test harness"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def subTest(mol,queries,qfps,fpf,verify,silent):\n",
      "    nFound=0\n",
      "    nScreened=0\n",
      "    nTot=0\n",
      "    nFailed = 0\n",
      "    molfp = fpf(mol,False)\n",
      "    for j,queryfp in enumerate(qfps):\n",
      "        nTot += 1 \n",
      "        if DataStructs.AllProbeBitsMatch(queryfp,molfp):\n",
      "            nScreened += 1\n",
      "            if mol.HasSubstructMatch(queries[j]):\n",
      "                nFound += 1\n",
      "        elif verify:\n",
      "            if mol.HasSubstructMatch(queries[j]):\n",
      "                nFailed += 1\n",
      "                if not silent:\n",
      "                    print('Failure: %s %s'%(Chem.MolToSmiles(mol,True),Chem.MolToSmiles(queries[j],True)))\n",
      "    return nFound,nScreened,nTot,nFailed\n",
      "\n",
      "def testScreenout(mols,queries,fpf,dview,verify=False,silent=False):\n",
      "    nFound=0\n",
      "    nScreened=0\n",
      "    nTot=0\n",
      "    nFailed = 0\n",
      "    #if not silent: print(\"Building query fingerprints\")\n",
      "    qfps = [fpf(m,True) for m in queries]\n",
      "    #if not silent: print(\"Running Queries\")\n",
      "    t = lambda x,qfps=qfps,queries=queries,fpf=fpf,verify=verify,silent=silent,subTest=subTest:subTest(x,queries,qfps,fpf,verify,silent)\n",
      "    res=dview.map_async(t,mols)\n",
      "    for entry in res:\n",
      "        nFound+=entry[0]\n",
      "        nScreened+=entry[1]\n",
      "        nTot+=entry[2]\n",
      "        nFailed+=entry[3]\n",
      "    if not silent:\n",
      "        accuracy = float(nFound)/nScreened\n",
      "        print(\"Found %d matches in %d searches with %d failures. Accuracy: %.3f\"%(nFound,nScreened,nFailed,accuracy))\n",
      "    return nTot,nScreened,nFound,nFailed\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 6
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Tests"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "methods = {\n",
      "   'Avalon-2K':lambda x,y: pyAvalonTools.GetAvalonFP(x,nBits=2048,isQuery=y,bitFlags=pyAvalonTools.avalonSSSBits),\n",
      "\n",
      "   'Pattern-1K':lambda x,y: Chem.PatternFingerprint(x,fpSize=1024),\n",
      "   'Pattern-2K':lambda x,y: Chem.PatternFingerprint(x,fpSize=2048),\n",
      "   'Pattern-4K':lambda x,y: Chem.PatternFingerprint(x,fpSize=4096),\n",
      "\n",
      "   'Layered': lambda x,y:Chem.LayeredFingerprint(x,layerFlags=Chem.LayeredFingerprint_substructLayers),\n",
      "   }"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 11
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "leadResults={}\n",
      "fragResults={}\n",
      "pieceResults={}\n",
      "for method,func in methods.iteritems():\n",
      "    print(\"----------------------------\")\n",
      "    print(\"Doing %s\"%method)\n",
      "    print(\"Leads\")\n",
      "    leadResults[method]=testScreenout(mols,leads,func,dview)\n",
      "    print(\"Frags\")\n",
      "    fragResults[method]=testScreenout(mols,frags,func,dview)\n",
      "    print(\"Pieces\")\n",
      "    pieceResults[method]=testScreenout(mols,pieces,func,dview)\n",
      "    "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "----------------------------\n",
        "Doing Avalon-2K\n",
        "Leads\n",
        "Found 886 matches in 2044 searches with 0 failures. Accuracy: 0.433"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Frags\n",
        "Found 4479 matches in 13656 searches with 0 failures. Accuracy: 0.328"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Pieces\n",
        "Found 1926728 matches in 4141259 searches with 0 failures. Accuracy: 0.465"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "----------------------------\n",
        "Doing Layered\n",
        "Leads\n",
        "Found 1274 matches in 6797 searches with 0 failures. Accuracy: 0.187"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Frags\n",
        "Found 4659 matches in 44258 searches with 0 failures. Accuracy: 0.105"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Pieces\n",
        "Found 1935315 matches in 4642461 searches with 0 failures. Accuracy: 0.417"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "----------------------------\n",
        "Doing Pattern-2K\n",
        "Leads\n",
        "Found 1274 matches in 1781 searches with 0 failures. Accuracy: 0.715"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Frags\n",
        "Found 4659 matches in 7892 searches with 0 failures. Accuracy: 0.590"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Pieces\n",
        "Found 1935315 matches in 3381436 searches with 0 failures. Accuracy: 0.572"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "----------------------------\n",
        "Doing Pattern-1K\n",
        "Leads\n",
        "Found 1274 matches in 7527 searches with 0 failures. Accuracy: 0.169"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Frags\n",
        "Found 4659 matches in 16372 searches with 0 failures. Accuracy: 0.285"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Pieces\n",
        "Found 1935315 matches in 3874208 searches with 0 failures. Accuracy: 0.500"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "----------------------------\n",
        "Doing Pattern-4K\n",
        "Leads\n",
        "Found 1274 matches in 1748 searches with 0 failures. Accuracy: 0.729"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Frags\n",
        "Found 4659 matches in 7742 searches with 0 failures. Accuracy: 0.602"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Pieces\n",
        "Found 1935315 matches in 3048867 searches with 0 failures. Accuracy: 0.635"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n"
       ]
      }
     ],
     "prompt_number": 12
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## Summarize that:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mns = sorted(methods.keys())\n",
      "for mn in mns:\n",
      "    print(mn,'Pieces','%.3f'%(float(pieceResults[mn][2])/pieceResults[mn][1]),'Fragments','%.3f'%(float(fragResults[mn][2])/fragResults[mn][1]),\n",
      "          'Leads','%.3f'%(float(leadResults[mn][2])/leadResults[mn][1]))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Avalon-2K Pieces 0.465 Fragments 0.328 Leads 0.433\n",
        "Layered Pieces 0.417 Fragments 0.105 Leads 0.187\n",
        "Pattern-1K Pieces 0.500 Fragments 0.285 Leads 0.169\n",
        "Pattern-2K Pieces 0.572 Fragments 0.590 Leads 0.715\n",
        "Pattern-4K Pieces 0.635 Fragments 0.602 Leads 0.729\n"
       ]
      }
     ],
     "prompt_number": 14
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## A caveat\n",
      "\n",
      "The results from the Avalon fingerprint shown above are not directly comparable to the others since structures are being filtered out that shouldn't be.\n",
      "\n",
      "To demonstrate this, we need a serial form of the screenout test:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def testScreenoutSerial(mols,queries,fpf,verify=False,silent=False):\n",
      "    nFound=0\n",
      "    nScreened=0\n",
      "    nTot=0\n",
      "    nFailed = 0\n",
      "    #if not silent: print(\"Building query fingerprints\")\n",
      "    qfps = [fpf(m,True) for m in queries]\n",
      "    #if not silent: print(\"Running Queries\")\n",
      "    t = lambda x,qfps=qfps,queries=queries,fpf=fpf,verify=verify,silent=silent,subTest=subTest:subTest(x,queries,qfps,fpf,verify,silent)\n",
      "    res=map(t,mols)\n",
      "    for entry in res:\n",
      "        nFound+=entry[0]\n",
      "        nScreened+=entry[1]\n",
      "        nTot+=entry[2]\n",
      "        nFailed+=entry[3]\n",
      "    if not silent:\n",
      "        accuracy = float(nFound)/nScreened\n",
      "        print(\"Found %d matches in %d searches with %d failures. Accuracy: %.3f\"%(nFound,nScreened,nFailed,accuracy))\n",
      "    return nTot,nScreened,nFound,nFailed\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 8
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "And now we can get some examples:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "testScreenoutSerial(mols[:500],leads,methods['Avalon-2K'],verify=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Failure: Cc1ccc(C)c(S(=O)(=O)c2nnn3c4ccsc4c(Nc4ccc(C)c(C)c4)nc23)c1 c1cn[nH]n1\n",
        "Failure: COc1ccc(OCC#Cc2cn([C@H](C)C[C@@H]3CC[C@H]([C@@H](C)C(=O)N(C)Cc4ccccc4)O3)nn2)cc1 c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: Cc1c(C(=O)Nc2ccccc2)nnn1Cc1ccccc1O c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: c1c(CCc2ccccc2)nnn1C[C@H]1CC[C@@H]([C@@H]2CC[C@H](Cn3cc(CCc4ccccc4)nn3)O2)O1 c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=C(NCc1ccc(F)cc1)[C@@H](CCN1[C@@H]2CC[C@@H]1CC(n1nnc3cccnc31)C2)c1ccccc1 c1cn[nH]n1\n",
        "Found 10 matches in 22 searches with 5 failures. Accuracy: 0.455\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 16,
       "text": [
        "(250000, 22, 10, 5)"
       ]
      }
     ],
     "prompt_number": 16
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Look at a specific example:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mfp = pyAvalonTools.GetAvalonFP('Cc1ccc(C)c(S(=O)(=O)c2nnn3c4ccsc4c(Nc4ccc(C)c(C)c4)nc23)c1',True,2048,isQuery=False)\n",
      "qfp = pyAvalonTools.GetAvalonFP('c1cn[nH]n1',True,2048,isQuery=True)\n",
      "DataStructs.AllProbeBitsMatch(qfp,mfp)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 17,
       "text": [
        "False"
       ]
      }
     ],
     "prompt_number": 17
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "Chem.MolFromSmiles('Cc1ccc(C)c(S(=O)(=O)c2nnn3c4ccsc4c(Nc4ccc(C)c(C)c4)nc23)c1')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAcIAAACWCAYAAABNcIgQAAAfE0lEQVR4nO3de1iVVb4H8C8iDCoI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\niYhIqzEIiYhIq/0fKXgSLh0ZenMAAAAASUVORK5CYII=\n",
       "prompt_number": 18,
       "text": [
        "<rdkit.Chem.rdchem.Mol at 0x5ca01360>"
       ]
      }
     ],
     "prompt_number": 18
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The normal explanation for this would be differences in the aromaticity model. In this case, however, that turns out not be exactly it.\n",
      "\n",
      "Here's the molecule we're querying with:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "Chem.MolFromSmiles('c1cn[nH]n1')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAcIAAACWCAYAAABNcIgQAAAMGUlEQVR4nO3de0zV9R/H8RcUolY2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\nb1pVsmEqIgAAAABJRU5ErkJggg==\n",
       "prompt_number": 19,
       "text": [
        "<rdkit.Chem.rdchem.Mol at 0x5ca80980>"
       ]
      }
     ],
     "prompt_number": 19
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "qfp2 = pyAvalonTools.GetAvalonFP('c1cnn[nH]1',True,2048,isQuery=True)\n",
      "qfp==qfp2"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 20,
       "text": [
        "False"
       ]
      }
     ],
     "prompt_number": 20
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The deciding factor is the tautomer. The Avalon fingerprinter considers this significant,\n",
      "and that difference explains the different screenout.\n",
      "\n",
      "If we use the fingerprint from the second tautomer we see the expected screenout behavior:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "DataStructs.AllProbeBitsMatch(qfp2,mfp)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 102,
       "text": [
        "True"
       ]
      }
     ],
     "prompt_number": 102
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "This ends up mattering because the RDKit's substructure matcher ignores the H completely:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "Chem.MolFromSmiles('c1cnn[nH]1').HasSubstructMatch(Chem.MolFromSmiles('c1cn[nH]n1'))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 103,
       "text": [
        "True"
       ]
      }
     ],
     "prompt_number": 103
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "Chem.MolFromSmiles('c1cnnn1C').HasSubstructMatch(Chem.MolFromSmiles('c1cn[nH]n1'))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 104,
       "text": [
        "True"
       ]
      }
     ],
     "prompt_number": 104
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Unless, of course, you construct the query from SMARTS:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "Chem.MolFromSmiles('c1cnnn1C').HasSubstructMatch(Chem.MolFromSmarts('c1cn[nH]n1'))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 105,
       "text": [
        "False"
       ]
      }
     ],
     "prompt_number": 105
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Other fingerprints"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## The RDKit fingerprint"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "fn = lambda x,v:Chem.RDKFingerprint(x,minPath=1,maxPath=5,nBitsPerHash=1,useHs=False)\n",
      "testScreenoutSerial(mols,leads,fn,verify=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Failure: COc1ccc(-n2c(SC(C)=O)nc3sc4c(c3c2=O)CCCC4)cc1 CSC(C)=O\n",
        "Failure: CCCCCC[C@H]1SC(=O)c2ccccc2[C@@H]1C(=O)O CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccccc1SC(=O)c1cccc(C=O)n1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(-n2c(SC(C)=O)nc3sc4c(c3c2=O)CCCC4)cc1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CC(C)(C)OC(=O)n1c(-c2ccc3c(c2)CC(NS(=O)(=O)c2ccccc2)C3)cc2cc(O)ccc21 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: Cc1c2cc(O)ccc2n(Cc2ccc(OC[C@H](C)N3CCCC3)cc2)c1-c1ccc(O)cc1 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CC1(C)Oc2ccc(-c3nc(-c4cccc5c(CCC(=O)O)c[nH]c54)no3)cc2O1 Cc1noc(-c2ccc(O)cc2)n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(CC(=O)Nc2ccc(-c3noc(-c4cc(OC)c(OC)c(OC)c4)n3)cc2)cc1 Cc1noc(-c2ccc(O)cc2)n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CO[C@H]1C(=O)N2C(C(=O)C(C)(C)C)=C(C)[C@@H](S(=O)(=O)c3ccccc3)S(=O)(=O)C12 CSCSC"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: Cc1c2c(nc3ccc(OCCN4CCCCC4)cc13)-c1cc3cc(O)ccc3n1C2 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc2[nH]c(-c3ccc4ccccc4c3)c(-c3cc(OC)c(OC)c(OC)c3)c2c1 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=C(O)C1CCCN1C(=O)CC(SC(=O)c1ccccc1)C(=O)c1ccccc1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CC(C)(C)OC(=O)n1c(-c2ccc3c(c2)CC(NS(=O)(=O)c2ccccc2)C3)cc2cc(O)ccc21 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=Cc1cccc(C(=O)Sc2ccc(Cl)cc2)n1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=C1NCc2cccc(-c3cc4cc(OCCN5CCCCC5)ccc4[nH]3)c21 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(C#CCC2(S(=O)(=O)c3ccc(C)cc3)SC(=O)NC2=O)cc1 CSCSC"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CCCC(=O)Sc1nc2sc3c(c2c(=O)n1-c1ccc(OC)cc1)CCCC3 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(-n2c(SC(C)=O)nc3sc4c(c3c2=O)CCCC4)cc1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(-c2nc(-c3ccc(NC(=O)c4ccc(Br)o4)cc3)no2)cc1OC Cc1noc(-c2ccc(O)cc2)n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1cc2c(cc1OC)C(=O)N1CSCC1C(=O)S2 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: Oc1nc2ccccc2cc1-c1cc2cc(OCCN3CCCCC3)ccc2[nH]1 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CC(C)(C)OC(=O)[C@@H]1CCCN1C(=O)[C@H]1CCC[C@@H]1SC(=O)c1ccccc1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=C1Sc2ccccc2C(=O)N2CCSC12 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CC(C)(C)OC(=O)n1c(-c2ccc3c(c2)CC(NS(=O)(=O)c2ccccc2)C3)cc2cc(O)ccc21 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=c1n(Cc2ccccc2)nc2n1-c1ccccc1OC2 CCn1nc2n(c1=O)-c1ccccc1OC2"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(-n2c(SC(C)=O)nc3sc4c(c3c2=O)CCCC4)cc1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CCCOC(=O)c1c(CCC)c(C(=O)SCC)c(CC)nc1-c1cccc(F)c1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=C1Sc2sc(=S)sc2SC1c1ccccc1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(-n2c(SC(C)=O)nc3sc4c(c3c2=O)CCCC4)cc1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc2c(c1)c(CCN(C)C)c1n2S(=O)(=O)c2ccccc2-1 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CC(C)(C)OC(=O)n1c(-c2ccc3c(c2)CC(NS(=O)(=O)c2ccccc2)C3)cc2cc(O)ccc21 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CCCC(=O)Sc1nc2sc3c(c2c(=O)n1-c1ccc(OC)cc1)CCCC3 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(-c2cc3cc(OC)ccc3[nH]2)cc1 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=C(O)[C@@H](Cc1ccc(-c2ccccc2)cc1)SC(=O)c1ccccc1 CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Found 1240 matches in 1399 searches with 34 failures. Accuracy: 0.886"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 27,
       "text": [
        "(25000000, 1399, 1240, 34)"
       ]
      }
     ],
     "prompt_number": 27
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The failures happen here because the RDKit fingerprinter uses information about aromaticity in the calculation of atom invariants and while hashing bonds. That extra information is a big part of why the accuracy is so high.\n",
      "\n",
      "Turning off all bond order information (finer grained control is unfortunately not possible) and using atomic number as the atom invariants gives no errors, but a much lower screenout rate:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def func(mol,v):\n",
      "    invs = [x.GetAtomicNum() for x in mol.GetAtoms()]\n",
      "    return Chem.RDKFingerprint(mol,minPath=1,maxPath=5,nBitsPerHash=1,useHs=False,useBondOrder=False,atomInvariants=invs)\n",
      "testScreenoutSerial(mols,leads,func,verify=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Found 1274 matches in 125651 searches with 0 failures. Accuracy: 0.010\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 53,
       "text": [
        "(25000000, 125651, 1274, 0)"
       ]
      }
     ],
     "prompt_number": 53
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Using a shorter maxPath length also hurts accuracy:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "fn = lambda x,v:Chem.RDKFingerprint(x,minPath=1,maxPath=4,nBitsPerHash=1,useHs=False)\n",
      "testScreenout(mols,leads,fn,dview)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Found 1240 matches in 2331 searches with 0 failures. Accuracy: 0.532\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 35,
       "text": [
        "(25000000, 2331, 1240, 0)"
       ]
      }
     ],
     "prompt_number": 35
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "It is most likely luck that there are no failures here. "
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## Chemfp\n",
      "\n",
      "Andrew Dalke's chemfp package (http://chemfp.com/) includes a pattern-based fingerprinter which is based on the PubChem/CACTVS substructure keys:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from chemfp import rdkit_patterns\n",
      "dalke_fp=rdkit_patterns.SubstructRDKitFingerprinter_v1.make_fingerprinter()\n",
      "fn = lambda x,v,fp=dalke_fp:DataStructs.CreateFromBinaryText(fp(x))\n",
      "testScreenoutSerial(mols[:500],leads,fn,verify=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Failure: CC(C)(C)c1cc2c(cc1SC1=C(O)OC(CCc3ccccc3)(c3ccccc3)CC1=O)CCCC2 C1=COCCC1\n",
        "Failure: CC(C)(C)c1cc2c(cc1SC1=C(O)OC(CCc3ccccc3)(c3ccccc3)CC1=O)CCCC2 O=C1C=CO[C@H](c2ccccc2)C1\n",
        "Failure: Cc1ccc(C)c(S(=O)(=O)c2nnn3c4ccsc4c(Nc4ccc(C)c(C)c4)nc23)c1 c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CCOC(=O)Cc1c(C)n(CC2CCCCC2)c2ccc(OC)cc12 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(OCC#Cc2cn([C@H](C)C[C@@H]3CC[C@H]([C@@H](C)C(=O)N(C)Cc4ccccc4)O3)nn2)cc1 c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COC1=C(OC)C(=O)C2=C(C[C@@H]3[C@@]4(C)C[C@H]4C[C@@]3(C)C2)C1=O COC1=CCC=CC1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc2[nH]c3c(c2c1)CC(NC(=O)C1CC1)CC3 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: Cc1c(C(=O)Nc2ccccc2)nnn1Cc1ccccc1O c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1c(=O)[nH]c2ccccc2c1-c1ccccc1 Oc1cnc2ccccc2c1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: Cc1ccc2c(nc(-c3ccc(NC(=O)C(F)(F)F)cc3)c(O)c2C(=O)O)c1C Oc1cnc2ccccc2c1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: c1c(CCc2ccccc2)nnn1C[C@H]1CC[C@@H]([C@@H]2CC[C@H](Cn3cc(CCc4ccccc4)nn3)O2)O1 c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=C(NCc1ccc(F)cc1)[C@@H](CCN1[C@@H]2CC[C@@H]1CC(n1nnc3cccnc31)C2)c1ccccc1 c1cn[nH]n1\n",
        "Found 3 matches in 108 searches with 12 failures. Accuracy: 0.028\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 31,
       "text": [
        "(250000, 108, 3, 12)"
       ]
      }
     ],
     "prompt_number": 31
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "It's not particularly effective as a screening fingerprint and, as seen above, creates some holes."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## MACCS keys\n",
      "\n",
      "Another key-based fingerprint that isn't particularly effective and results in some misses:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "fn = lambda x,v:rdMolDescriptors.GetMACCSKeysFingerprint(x)\n",
      "testScreenoutSerial(mols[:500],leads,fn,verify=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Failure: CC(C)(C)c1cc2c(cc1SC1=C(O)OC(CCc3ccccc3)(c3ccccc3)CC1=O)CCCC2 C1=COCCC1\n",
        "Failure: Cc1ccc(C)c(S(=O)(=O)c2nnn3c4ccsc4c(Nc4ccc(C)c(C)c4)nc23)c1 c1cn[nH]n1\n",
        "Failure: CCOC(=O)C1=COC(C)(CCc2ccccc2)CC1=O C1=COCCC1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CCOC(=O)Cc1c(C)n(CC2CCCCC2)c2ccc(OC)cc12 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc(OCC#Cc2cn([C@H](C)C[C@@H]3CC[C@H]([C@@H](C)C(=O)N(C)Cc4ccccc4)O3)nn2)cc1 c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COC1=C(OC)C(=O)C2=C(C[C@@H]3[C@@]4(C)C[C@H]4C[C@@]3(C)C2)C1=O COC1=CCC=CC1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1ccc2[nH]c3c(c2c1)CC(NC(=O)C1CC1)CC3 Cc1cc2cc(O)ccc2[nH]1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: Cc1c(C(=O)Nc2ccccc2)nnn1Cc1ccccc1O c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: COc1c(=O)[nH]c2ccccc2c1-c1ccccc1 Oc1cnc2ccccc2c1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: CC(=O)NCCCC[C@H](NC(C)=O)C(=O)NCC(=O)S[C@H](C)C(=O)O CSC(C)=O"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: c1c(CCc2ccccc2)nnn1C[C@H]1CC[C@@H]([C@@H]2CC[C@H](Cn3cc(CCc4ccccc4)nn3)O2)O1 c1cn[nH]n1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Failure: O=C(NCc1ccc(F)cc1)[C@@H](CCN1[C@@H]2CC[C@@H]1CC(n1nnc3cccnc31)C2)c1ccccc1 c1cn[nH]n1\n",
        "Found 3 matches in 134 searches with 12 failures. Accuracy: 0.022\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 9,
       "text": [
        "(250000, 134, 3, 12)"
       ]
      }
     ],
     "prompt_number": 9
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Improving things\n",
      "\n",
      "A first pass at improving is to combine some selected patterns with an algorithmic fingerprinter like the pattern fingerprinter. Here's a thread describing an experiment:\n",
      "\n",
      "http://www.mail-archive.com/rdkit-discuss@lists.sourceforge.net/msg02078.html\n",
      "\n",
      "And a reproduction of that with this test harness:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "qs=\"\"\"0 2 times O largest 55458\n",
      "1 2 times Ccc largest 29602\n",
      "2 1 times CCN largest 16829\n",
      "3 1 times cnc largest 11439\n",
      "4 1 times cN largest 8998\n",
      "5 1 times C=O largest 7358\n",
      "6 1 times CCC largest 6250\n",
      "7 1 times S largest 4760\n",
      "8 1 times c1ccccc1 largest 4524\n",
      "9 2 times N largest 2854\n",
      "10 1 times C=C largest 2162\n",
      "11 1 times nn largest 1840\n",
      "12 2 times CO largest 1248\n",
      "13 1 times Ccn largest 964\n",
      "14 1 times CCCCC largest 857\n",
      "15 1 times cc(c)c largest 653\n",
      "16 3 times O largest 653\n",
      "17 1 times O largest 466\n",
      "18 2 times CNC largest 464\n",
      "19 1 times s largest 457\n",
      "20 1 times CC(C)C largest 335\n",
      "21 1 times o largest 334\n",
      "22 1 times cncnc largest 334\n",
      "23 1 times C=N largest 321\n",
      "24 2 times CC=O largest 238\n",
      "25 4 times Ccc largest 238\n",
      "26 1 times Cl largest 230\n",
      "27 4 times O largest 149\n",
      "28 2 times ccncc largest 149\n",
      "29 6 times CCCCCC largest 76\n",
      "30 2 times c1ccccc1 largest 76\n",
      "31 1 times F largest 75\n",
      "32 3 times CCOC largest 44\n",
      "33 3 times N largest 44\n",
      "34 1 times c(cn)n largest 44\n",
      "35 1 times N largest 41\n",
      "36 9 times C largest 41\n",
      "37 1 times CC=C(C)C largest 33\n",
      "38 1 times c1ccncc1 largest 26\n",
      "39 1 times CC(C)N largest 26\n",
      "40 1 times CC largest 26\n",
      "41 4 times CCC(C)O largest 25\n",
      "42 2 times ccc(cc)n largest 21\n",
      "43 6 times C largest 21\n",
      "44 1 times C1CCCC1 largest 18\n",
      "45 1 times C largest 18\n",
      "46 5 times O largest 18\n",
      "47 2 times Ccn largest 14\n",
      "48 1 times CNCN largest 13\n",
      "49 3 times cncn largest 13\n",
      "50 1 times CSC largest 13\n",
      "51 3 times CC=O largest 11\n",
      "52 1 times CCNCCCN largest 11\n",
      "53 1 times CccC largest 11\n",
      "54 3 times ccccc(c)c largest 10\"\"\"\n",
      "\n",
      "def _initPatts():\n",
      "    ssPatts=[]\n",
      "    for q in qs.split('\\n'):\n",
      "        q = q.split(' ')\n",
      "        count = int(q[1])\n",
      "        patt = Chem.MolFromSmiles(q[3],sanitize=False)\n",
      "        patt.UpdatePropertyCache(strict=False) # <- github #149, without this we cannot pickle the queries\n",
      "        ssPatts.append((patt,count))\n",
      "    return ssPatts\n",
      "    \n",
      "_ssPatts=None\n",
      "def GetCombinedSubstructFP(m,ssPatts=None,fpSize=1024,verbose=False):\n",
      "    if ssPatts is None:\n",
      "        ssPatts = _initPatts()\n",
      "    sz = len(ssPatts)\n",
      "    lfp=Chem.PatternFingerprint(m,fpSize=fpSize)\n",
      "    res = DataStructs.ExplicitBitVect(fpSize+sz)\n",
      "    obls = [x+sz for x in lfp.GetOnBits()]\n",
      "    res.SetBitsFromList(obls)\n",
      "    for i,(p,count) in enumerate(ssPatts):\n",
      "        matches = m.GetSubstructMatches(p,uniquify=True)\n",
      "        if len(matches)>=count:\n",
      "            res.SetBit(i)\n",
      "    return res\n",
      "            \n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 15
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "ssPatts = _initPatts()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 16
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "fn = lambda x,v,ssPatts=ssPatts:GetCombinedSubstructFP(x,ssPatts=ssPatts,fpSize=2048)\n",
      "testScreenoutSerial(mols[:500],leads,fn,verify=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Found 15 matches in 23 searches with 0 failures. Accuracy: 0.652\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 65,
       "text": [
        "(250000, 23, 15, 0)"
       ]
      }
     ],
     "prompt_number": 65
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "leadResults={}\n",
      "fragResults={}\n",
      "pieceResults={}\n",
      "print(\"----------------------------\")\n",
      "fn = lambda x,v,ssPatts=ssPatts,fpf=GetCombinedSubstructFP:fpf(x,ssPatts=ssPatts,fpSize=1024)\n",
      "method='1024'\n",
      "print(\"Doing %s\"%method)\n",
      "print(\"Pieces\")\n",
      "pieceResults[method]=testScreenout(mols,pieces,fn,dview)\n",
      "print(\"Frags\")\n",
      "fragResults[method]=testScreenout(mols,frags,fn,dview)\n",
      "print(\"Leads\")\n",
      "leadResults[method]=testScreenout(mols,leads,fn,dview)\n",
      "\n",
      "print(\"----------------------------\")\n",
      "fn = lambda x,v,ssPatts=ssPatts,fpf=GetCombinedSubstructFP:fpf(x,ssPatts=ssPatts,fpSize=2048)\n",
      "method='2048'\n",
      "print(\"Doing %s\"%method)\n",
      "print(\"Pieces\")\n",
      "pieceResults[method]=testScreenout(mols,pieces,fn,dview)\n",
      "print(\"Frags\")\n",
      "fragResults[method]=testScreenout(mols,frags,fn,dview)\n",
      "print(\"Leads\")\n",
      "leadResults[method]=testScreenout(mols,leads,fn,dview)\n",
      "      \n",
      "      \n",
      "      "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "----------------------------\n",
        "Doing 1024\n",
        "Pieces\n",
        "Found 1935315 matches in 3023990 searches with 0 failures. Accuracy: 0.640"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Frags\n",
        "Found 4659 matches in 13765 searches with 0 failures. Accuracy: 0.338"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Leads\n",
        "Found 1274 matches in 6857 searches with 0 failures. Accuracy: 0.186"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "----------------------------\n",
        "Doing 2048\n",
        "Pieces\n",
        "Found 1935315 matches in 2810633 searches with 0 failures. Accuracy: 0.689"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Frags\n",
        "Found 4659 matches in 7669 searches with 0 failures. Accuracy: 0.608"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Leads\n",
        "Found 1274 matches in 1761 searches with 0 failures. Accuracy: 0.723"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n"
       ]
      }
     ],
     "prompt_number": 17
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Comparing to the previous pattern fp results:\n",
      "```\n",
      "Pattern-1K Pieces 0.500 Fragments 0.285 Leads 0.169\n",
      "Combine-1K Pieces 0.640 Fragments 0.338 Leads 0.186\n",
      "Pattern-2K Pieces 0.572 Fragments 0.590 Leads 0.715\n",
      "Combine-2K Pieces 0.689 Fragments 0.608 Leads 0.723\n",
      "Pattern-4K Pieces 0.635 Fragments 0.602 Leads 0.729\n",
      "\n",
      "```\n",
      "These help a bit, particularly with the pieces, but not a whole lot.\n"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    }
   ],
   "metadata": {}
  }
 ]
}