a#nSc @sdZddlZddlZddlZddlmZddlmZddlmZddl m Z ddl Z ddl m Z ddlZdZejd d Zejd Zd eee!Zejd ZddddddZddddddZddddddZddddddddZdddddddZddZddd dZdejdd ddZdZ dZ!de"fdYZ#d Z$e%d!krddl&Z&e#Z'e$e'dd"l(m)Z)e e'Z*dZ+dZ,e'j-re*sAed#ej.d$ned%e&j&Z/ee*d&e'j0d'e'j1d(e'j2Z+e&j&Z3ed)e3e/e'j4reed*ej5e+e6e'j4d+qene'j4reed,ej7e6e'j4d-Z+e'j8reyej7e6e'j8d-Z,Wn dZ,qbXe9e,e:j;d.d krbdZ,qbqendZ<e'j=re*sed/ej.d$ne+sed0ej.d$ned1e, se9e,e'j0kr]ee*e+d&e'j0d2e'j>d3e'j?\Z<Z,e'j@r2ej5e<e6e'j@d+ne'j8rej5e,e6e'j8d+qqee,e*e+d&e'j0d2e'j>d3e'j?Z<n'e'j@rej7e6e'j@d-Z<ne'jArejBjCe'jAre'jD red4ej7e6e'jAd-ZEngZEdZFe'jGre*s9ed5ej.d$ne+pBeEs_ed6ej.d$ned7e&j&Z/e'jArd ZHndZHeEsee*e+d8e'jId2e'j>d3e'j?d9e'jJd:eH\ZFZEe'jArMed;e6e'jAd+ZKej5eEeKd eKjLqMn3ee*eEd8e'jId2e'j>d3e'j?d9e'jJZFe&j&Z3ed)e3e/e'jMre6e'jMd<ZNeeNeFe+d3e'j?qn*e'jMre6e'jMd=ZOeeOZFne'jPre+sed>ej.d$neFsed?ej.d$ne)ZQeQjRd@ee+eFdAe'jSdBeQe'jTrue6e'jTd<jReQjUqejVjReQjUqndS(Cse command line utility for working with FragmentCatalogs (CASE-type analysis) **Usage** BuildFragmentCatalog [optional args] filename, the name of a delimited text file containing InData, is required for some modes of operation (see below) **Command Line Arguments** - -n *maxNumMols*: specify the maximum number of molecules to be processed - -b: build the catalog and OnBitLists *requires InData* - -s: score compounds *requires InData and a Catalog, can use OnBitLists* - -g: calculate info gains *requires Scores* - -d: show details about high-ranking fragments *requires a Catalog and Gains* - --catalog=*filename*: filename with the pickled catalog. If -b is provided, this file will be overwritten. - --onbits=*filename*: filename to hold the pickled OnBitLists. If -b is provided, this file will be overwritten - --scores=*filename*: filename to hold the text score data. If -s is provided, this file will be overwritten - --gains=*filename*: filename to hold the text gains data. If -g is provided, this file will be overwritten - --details=*filename*: filename to hold the text details data. If -d is provided, this file will be overwritten. - --minPath=2: specify the minimum length for a path - --maxPath=6: specify the maximum length for a path - --smiCol=1: specify which column in the input data file contains SMILES - --actCol=-1: specify which column in the input data file contains activities - --nActs=2: specify the number of possible activity values - --nBits=-1: specify the maximum number of bits to show details for iN(tChem(tRDConfig(tFragmentCatalog(t DbConnect(t InfoTheorys $Revision$t:it$s%scCs|j|dS(N(twrite(tmsgtdest((sa/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/python2.7/site-packages/rdkit/Chem/BuildFragmentCatalog.pytmessagePsiii c Cs+|dkr'tjjtjd}ntj|||}tj|}tj }|dkrl|} n$t |drt |} nd} xt |D]\} } | | krPn| r| | r| dkrt d| | |jfqt d| |jfn|j| |qW|S(sD builds a fragment catalog from a set of molecules in a delimited text block **Arguments** - suppl: a mol supplier - maxPts: (optional) if provided, this will set an upper bound on the number of points to be considered - groupFileName: (optional) name of the file containing functional group information - minPath, maxPath: (optional) names of the minimum and maximum path lengths to be considered - reportFreq: (optional) how often to display status information **Returns** a FragmentCatalog sFunctionalGroups.txtit__len__isDone %d of %d, %d paths sDone %d, %d paths N(tNonetostpathtjoinRt RDDataDirRt FragCatParamst FragCatalogtFragCatGeneratorthasattrtlent enumerateR t GetFPLengthtAddFragsFromMol( tsuppltmaxPtst groupFileNametminPathtmaxPatht reportFreqtfpParamstcatalogtfgentnPtstitmol((sa/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/python2.7/site-packages/rdkit/Chem/BuildFragmentCatalog.pyt BuildCatalogSs&       tcCs|j}tj|d|ftj}g} | rU| rU|djd}ntj} |jd} x7|D]/} | r| | rtd| n| r|st| j |} n|| d} | j | |}| j g|j D] }|^qx-t |D]}||d| fcd7ddl}ye|jtjdddddddd d d d d dddddddddddg\}}Wn4tjjdddl}|jtnXxS|D]K\}}|dkrt||_ q|dkrd|_ q|dkr d|_ q|dkr!d |_ q|d!kr9||_ q|d"krQd|_ q|d#krid|_ q|d$krd|_q|d%krd|_q|d&krd|_q|d'kr||_q|d(kr||_q|d)kr||_q|d*kr||_q|d+kr)||_q|d,krA||_q|d-kr_t||_q|d.kr}t||_q|d/kryt||_Wqtk r||_qXq|d0kryt||_Wqtk r||_qXq|d1kr:yt||_Wqtk r6||_qXq|d2krXt||_q|d3krvt||_q|d4krd5|_q|d6krt t!||_"q|d7krt||_#q|d8krttj$d5qtqWt%|r3|j r#|d5|_&q:|d5|_'ntdS(9Niisn:d:cstscatalog=sonbits=s scoresFile=s gainsFile=s detailsFile=sfpFile=sminPath=smaxPath=ssmiCol=sactCol=snameCol=snActs=snBits=s biasList=stopN=tbuildtsigsRQRtscoretnoTitlesError parsing command line: s-ns-cRWs-st s-ts s-ds--builds--scores--gainss--sigss-detailss --catalogs--onbitss --scoresFiles --gainsFiles --detailsFiles--fpFiles --minPaths --maxPaths--smiCols--actCols --nameCols--nActss--nBitss --noTitleis --biasLists--topNs-h((tgetoptRtargvtstderrRt tracebackt print_excRR)RRfRzRRRRRRRRRRRRRRt ValueErrorRR~R4R5RttupletevalRORNRRR{R}(RRtargstextrasRtargtval((sa/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/python2.7/site-packages/rdkit/Chem/BuildFragmentCatalog.pyt ParseArgss   %                                                    t__main__(tStringIOs(We require inData to generate a catalog isBuilding catalog RRRs That took %.2f seconds. sDumping catalog data swb+sLoading catalog trbs s%We require inData to score molecules s(We require a catalog to score molecules sScoring compounds R2R4s Reading fingerprints from file. s%We require inData to calculate gains s?We require either a catalog or fingerprints to calculate gains sCalculating Gains RNRORPsWriting fingerprint file. sw+trs$We require a catalog to get details s%We require gains data to get details sid,SMILES,gain RoRY(Wt__doc__RR tcPickletrdkitRRt rdkit.ChemRtrdkit.Dbase.DbConnectionRR'trdkit.MLRRt _cvsVersiontfindtidx1trfindtidx2RtstdoutR R R%R>RDRURVR_RmRrRRtobjectRRRttimeRt cStringIORRRZR7RRtt1RRRtt2RtdumptopentloadRRtinDtcounttscoresRRR4RRRtexistsRRTRQRRPRNROttmpFtcloseRRYReRtioRR5RtgetvalueR(((sa/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/python2.7/site-packages/rdkit/Chem/BuildFragmentCatalog.pytAs $  / > 7 4 *    %  X            "           "  (                    "