x Yc@s2dZddlZddlZddlZddlZddlmZddlmZm Z ddl m Z ddl m Z mZmZmZmZmZmZmZmZmZidd6d d 6d d 6d d6ZdZdZdZdZdZdZdZdZdZ dZ!dS(sQModule Description Copyright (c) 2010,2011 Tao Liu This code is free software; you can redistribute it and/or modify it under the terms of the BSD License (see the file COPYING included with the distribution). @status: experimental @version: $Revision$ @author: Tao Liu @contact: taoliu@jimmy.harvard.edu iN(t ArgumentError(tPopentPIPE(tlog( t BEDParsertELANDResultParsertELANDMultiParsertELANDExportParsert SAMParsert BAMParsert BAMPEParsert BEDPEParsert BowtieParsert guess_parserg`Athsg{Atmmg*uAtceg8Atdmc Csyt|j|_Wnhyt|j|_Wqtjd|jtjddjtjtjdqXnXt |_ |j j |_ |j dkrt |_n8|j dkrt|_n|j dkrt|_n|j dkrt|_n|j d kr!t|_n|j d krEt|_t|_ n|j d krrt|_t|_ t|_n{|j d krt|_t|_nW|j d krt|_n<|j dkrt|_n!tjd|j tjd|jdkr:|jdkr:|jjs:tjdtjdq:n|jdkrftjdtjdn|jr|jrtjdtjdn|jrdM|_!t"|jdd|_#n"t"|j$dd|_!dM|_#|jr t"|j%dd|_&n|j j |_ |j'd|_(|j'd|_)|j(|j)krrtjd|j'tjdnt*j+j|j,|j-d|_.t*j+j|j,|j-d|_/t*j+j|j,|j-d|_0t*j+j|j,|j-d|_1t*j+j|j,|j-d|_2t*j+j|j,|j-d|_3t*j+j|j,|j-d|_4t*j+j|j,|j-d |_5|j6rt*j+j|j,|j-d!|_7n dM|_7t*j+j|j,|j-d"|_8tj9d#d$|j:dd%d&d'd(d)tj;d*d+tj<|_tj=|_>tj?|_?tj@|_@d,jd-d.jtjAdd/d0|j-d1|j d2|jBd3|jCd4|jd5|jDd6|j'f |_E|jr+|jr|jEd7|j7_E|jEd8|j%7_E|jEd97_Eqy|jEd:|j7_E|jEd97_EnN|jrc|jEd;|j$7_E|jEd<|j%7_En|jEd=|j$7_E|jFr|jEd>7_E|jGdkr|jEd?|jG7_Eqn*|jHr|jEd@7_En|jEdA7_E|jIdBkr|jEdC|jI7_En|jCr9|jEdD|jJ|jKf7_En|jEdE|jK7_E|jrj|jEdF7_En|jEdG7_E|jLdBkr|jEdH|jL7_En|j dNkrd|_M|jEdI7_En|jEdJ7_E|jr|jEdK7_En|jNr|jOr|jEdL7_En|S(OsRValidate options from a OptParser object. Ret: Validated options object. s*Error when interpreting --gsize option: %ss4Available shortcuts of effective genome sizes are %st,itELANDtBEDt ELANDMULTIt ELANDEXPORTtSAMtBAMtBAMPEtBEDPEtBOWTIEtAUTOs!Format "%s" cannot be recognized!tautotalls1--keep-dup should be 'auto', 'all' or an integer!s--extsize must >= 1!s.--broad can't be combined with --call-summits!i iis8Upper limit of mfold should be greater than lower limit!s _peaks.xlss _peaks.beds_peaks.narrowPeaks_peaks.broadPeaks_peaks.gappedPeaks _summits.beds_treat_pileup.bdgs_control_lambda.bdgs_cutoff_analysis.txts_model.rtlevelitformats+%(levelname)-5s @ %(asctime)s: %(message)s tdatefmts%a, %d %b %Y %H:%M:%Ststreamtfilemodetws s# Command line: %st s# ARGUMENTS LIST:s # name = %ss # format = %ss# ChIP-seq file = %ss# control file = %ss# effective genome size = %.2es# band width = %ds# model fold = %s s1# pvalue cutoff for narrow/strong regions = %.2e s.# pvalue cutoff for broad/weak regions = %.2e sH# qvalue will not be calculated and reported as -1 in the final output. s# pvalue cutoff = %.2e s1# qvalue cutoff for narrow/strong regions = %.2e s.# qvalue cutoff for broad/weak regions = %.2e s# qvalue cutoff = %.2e sC# Larger dataset will be randomly sampled towards smaller dataset. s"# Random seed has been set as: %d s9# Smaller dataset will be scaled towards larger dataset. s9# Larger dataset will be scaled towards smaller dataset. g?s&# Using a custom scaling factor: %.2e s># Range for calculating regional lambda is: %d bps and %d bps s3# Range for calculating regional lambda is: %d bps s# Broad region calling is on s# Broad region calling is off s0# Additional cutoff on fold-enrichment is: %.2f s# Paired-End mode is on s# Paired-End mode is off s&# Searching for subpeak summits is on s:# MACS will save fragment pileup signal per million reads N(RR(PtefgsizetgsizetfloattloggingterrortjointkeystsystexittFalset gzip_flagR tupperRtparserRRRRR tTrueR tnomodelR R R tkeepduplicatestisdigittextsizetbroadt call_summitstpvaluetNonet log_qvalueRt log_pvaluetqvaluet broadcutofftlog_broadcutofftmfoldtlmfoldtumfoldtostpathtoutdirtnametpeakxlstpeakbedtpeakNarrowPeakt peakBroadPeaktpeakGappedPeakt summitbedt bdg_treatt bdg_controltcutoff_analysistcutoff_analysis_filetmodelRt basicConfigtverbosetstderrtcriticaltwarningtwarntdebugtinfotargvttfiletcfiletbwtargtxtt downsampletseedttolargetratiot smalllocalt largelocaltfecutofftshifttdo_SPMRt store_bdg(toptions((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pyt opt_validate*s                        """""""" % "                 "   c CsRt|_|jr@d&|_t|jdd|_d|_n+t|jdd|_d&|_d|_|j rd&|_ t|j dd|_ d|_ n+t|j dd|_ d&|_ d|_ |jd|_|jd|_|jd|_|jd|_|jd |_|jd |_|jd |_t|_|jd kpa|jd ks|jd krtd d n!|jd krtd d nt|_djdd|jd|jd|jd|jd|jd|jd|jd f |_ ndjdd|jd|jd|jd|jd|jd f|_ |jr|j d|j7_ |j d7_ n|j d|j7_ |j r|j d|j 7_ |j d7_ n|j d|j 7_ t!j"dd|j#ddd d!d"d#t$j%d$d%t!j&|_'t!j(|_)t!j*|_*t!j+|_+|S('sRValidate options from a OptParser object. Ret: Validated options object. i itptqs_diffpeaks.xlss_diffpeaks.beds_diffpeaks_by_peaks1.xlss_diffpeaks_by_peaks2.xlss _logLR.bdgttpeaks1sEMust specify both peaks1 and peaks2, or neither (to call peaks again)tpeaks2s s# ARGUMENTS LIST:s # name = %ss# ChIP-seq file 1 = %ss# control file 1 = %ss# ChIP-seq file 2 = %ss# control file 2 = %ss# Peaks, condition 1 = %ss# Peaks, condition 2 = %ss-# treat/control -log10(pvalue) cutoff = %.2e s^# treat/control -log10(qvalue) will not be calculated and reported as -1 in the final output. s-# treat/control -log10(qvalue) cutoff = %.2e s$# differential pvalue cutoff = %.2e sU# differential qvalue will not be calculated and reported as -1 in the final output. s$# differential qvalue cutoff = %.2e RiR s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$N(,R/R0t peaks_pvalueR;tpeaks_log_qvalueRtpeaks_log_pvaluettrack_score_methodt peaks_qvaluet diff_pvalueR<R=t score_methodt diff_qvalueRGRHRItpeak1xlstpeak2xlstbdglogLRt bdgpvaluetbdglogFCR3t call_peaksRoRpRR+tt1bdgtc1bdgtt2bdgtc2bdgR_R)RSRTR-RURVR*RWRXRYRZ(Rj((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytdiff_opt_validates                               c Cseyt|j|_Wnhyt|j|_Wqtjd|jtjddjtjtjdqXnXt |_ |j j |_ |j dkrt |_n|j dkrt|_n|j dkrt|_n|j dkrt|_n|j d kr!t|_n{|j d krEt|_t|_ nW|j d kr`t|_n<|j d kr{t|_n!tjd |j tjd|jdkr|jdkr|jjstjdtjdqn|j j |_ tjdd|jddddddtjddtj|_tj|_tj|_tj |_ |S(sRValidate options from a OptParser object. Ret: Validated options object. s*Error when interpreting --gsize option: %ss4Available shortcuts of effective genome sizes are %sRiRRRRRRRRs!Format "%s" cannot be recognized!RRs1--keep-dup should be 'auto', 'all' or an integer!Rii R s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$(!R&R'R(R)R*R+R,R-R.R/R0R R1RR2RRRRR R3R R R5R6RSRTRURVRWRXRYRZ(Rj((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytopt_validate_filterdupTsZ                  c Cs t|_|jj|_|jdkr6t|_n|jdkrQt|_n|jdkrlt|_n|jdkrt|_n|jdkrt |_n{|jdkrt |_t |_nW|jdkrt |_n<|jdkrt |_n!tjd |jtjd |jj|_|jrg|jd krtjd tjd qn8|jr|jd krtjdtjd qntjdd|jddddddtjddtj|_tj|_tj|_tj|_|S(sRValidate options from a OptParser object. Ret: Validated options object. RRRRRRRRs!Format "%s" cannot be recognized!igY@sKPercentage can't be bigger than 100.0. Please check your options and retry!isXNumber of tags can't be smaller than or equal to 0. Please check your options and retry!Rii R s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$(R/R0R R1RR2RRRRR R3R R R)R*R-R.t percentagetnumberRSRTRURVRWRXRYRZ(Rj((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytopt_validate_randsamplesP                    c Cst|_|jj|_|jdkr6t|_n|jdkrQt|_n|jdkrlt|_n|jdkrt|_n|jdkrt |_n{|jdkrt |_t |_nW|jdkrt |_n<|jdkrt |_n!tjd |jtjd |jj|_tjd d |jd dddddtjddtj|_tj|_tj|_tj|_|S(sRValidate options from a OptParser object. Ret: Validated options object. RRRRRRRRs!Format "%s" cannot be recognized!iRii R s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$(R/R0R R1RR2RRRRR R3R R R)R*R-R.RSRTRURVRWRXRYRZ(Rj((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytopt_validate_refinepeaks@                c Csyt|j|_Wnhyt|j|_Wqtjd|jtjddjtjtjdqXnXt |_ |j j |_ |j dkrt |_n8|j dkrt|_n|j dkrt|_n|j dkrt|_n|j d kr!t|_n|j d krEt|_t|_ n|j d krrt|_t|_ t|_n{|j d krt|_t|_nW|j d krt|_n<|j dkrt|_n!tjd|j tjd|j j |_ |jd|_|jd|_|j|jkrUtjd|jtjdntjj|j|j |_!tj"dd|j#ddddddtj$ddtj%|_tj&|_'tj(|_(tj)|_)|S(sRValidate options from a OptParser object. Ret: Validated options object. s*Error when interpreting --gsize option: %ss4Available shortcuts of effective genome sizes are %sRiRRRRRRRRRRs!Format "%s" cannot be recognized!is8Upper limit of mfold should be greater than lower limit!Rii R s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$(*R&R'R(R)R*R+R,R-R.R/R0R R1RR2RRRRR R3R R4R R R RARBRCRDRERFtrfileRRRSRTRURVRWRXRYRZ(Rj((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytopt_validate_predictdsl                      c Cst|_|jj|_|jdkr6t|_n|jdkrQt|_n|jdkrlt|_n|jdkrt|_n|jdkrt |_n{|jdkrt |_t |_nW|jdkrt |_n<|jdkrt |_n!tjd |jtjd |jj|_tjd d |jd dddddtjddtj|_tj|_tj|_tj|_|jdkrtjdtjd n|S(sRValidate options from a OptParser object. Ret: Validated options object. RRRRRRRRs!Format "%s" cannot be recognized!iRii R s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$is--extsize must > 0!(R/R0R R1RR2RRRRR R3R R R)R*R-R.RSRTRURVRWRXRYRZR7(Rj((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytopt_validate_pileupNsF                 c Cstjdddddddtjdd tj|_tj|_tj|_tj |_ xDt |j D]3}|dkrktjd|tj dqkqkW|j rt|j t|j krtjdtj dqn|S(sRValidate options from a OptParser object. Ret: Validated options object. RiR s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$tppoistqpoistsubtracttlogFEtFEtlogLRtslogLRtmaxsInvalid method: %sisMThe number and the order of arguments for --ofile must be the same as for -m.(RRRRRRRR(R)RSR-RURVR*RWRXRYRZtsettmethodR.tofiletlen(RjR((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytopt_validate_bdgcmps$        c Cstjdddddddtjdd tj|_tj|_tj|_tj |_ |j dkrtjd |j tj dnt |j dkrtjdtj dn|S(sRValidate options from a OptParser object. Ret: Validated options object. RiR s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$tfisherRtmeansInvalid method: %siisOnly support two replicates!(RRR(R)RSR-RURVR*RWRXRYRZRR.Rtifile(Rj((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytopt_validate_cmbrepss       c Cstjdddddddtjdd tj|_tj|_tj|_tj |_ |j j dkrtjd|j tj dn|j j dkr|j rtjdtj dn|S(sRValidate options from a OptParser object. Ret: Validated options object. RiR s+%(levelname)-5s @ %(asctime)s: %(message)s R!s%a, %d %b %Y %H:%M:%SR"R#R$tmultiplytaddtp2qRtminsInvalid method: %sis+Need EXTRAPARAM for method multiply or add!(RRRRR(RR(R)RSR-RURVR*RWRXRYRZRtlowerR.t extraparam(Rj((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pytopt_validate_bdgopts       ("t__doc__R-RDtreR)targparseRt subprocessRRtmathRtMACS2.IO.ParserRRRRRR R R R R R&RkRRRRRRRRR(((sP/oak/stanford/groups/akundaje/marinovg/programs/MACS-2.1.0/MACS2/OptValidator.pyts,    F   h B ; 1 L 6 # -