nQc@sdZddlZddlZddlZddlmZejdZdefdYZde fdYZ d e fd YZ d e fd YZ d Z dS(sP Represent a read alignment in SAM format Created on Sep 25, 2010 @author: ian iN(t AlignedReads(M|N|D|I|S|H|P)t SAMHeadercBsteZdZdZdZdZdZdZdZe dZ e dZ e d Z RS( cCswtj|tj|didd6tj|dgtj|dgtj|dgtj|dgdS(NtHDs1.3tVNtSQtRGtPGtCO(tdictt__init__t __setitem__(tself((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyR s  cCs6|dkr"tj|||ntd|dS(NRRRRRs&SAMHeader.__setitem__, Illegal key: %s(sHDsSQsRGsPGsCO(RR tKeyError(R tkeytvalue((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyR s cCs.|jdji|d6t|d6dS(NRtSNtLN(tgettappendtint(R tref_namet ref_length((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt add_refseqscCs4x-t||D]\}}|j||qWdS(N(tzipR(R t name_listt length_listRR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytpopulate_refseqs"scCsJd}x=t|jdD]&\}}|d|kr|}PqqW|S(NiRR(t enumerateR(R Rttidtitref((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytfind_tid&s "cCs|jd|jdS(NRR(R(R R((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt get_ref_name.sc Cs@g}d|jkrj|jddjg|jdjD]"\}}d|t|f^q:nxd D]}||jkrqxi|j|D]U}|jd|djg|jD]"\}}d|t|f^qfqWqqqqWd |jkr3x+|jd D]}|jd |qWnd j|S( NRs@HD %ss s%s:%sRRRs@%s %sRs@CO %ss (sSQsRGsPG(RRtjoinRtitemststr(R tlinesR Rtsuperkeytsubdicttitem((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt__str__1sU ZcCsTt}xD|D]<}|jj}|d}|d}|j||qW|S(Nii(RtstriptsplitR(R$tresulttlinetfieldsRR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt from_lines?s    cCs)t|}tj|}|j|S(N(topenRR.tclose(t fai_filenamethandleR+((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytfrom_faiIs  cCst}|j||S(N(Rtupdate(theaderR+((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytfrom_samfile_headerPs  ( t__name__t __module__R R RRRR R(t staticmethodR.R3R6(((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyRs        tSAMReadcBseZd8dZdZdZdZdZdZdZ dZ dZ d Z d Z d Zd Zd ZdZdZdZdZdZdZdZdZedZedZdZdZdZdZdZ dZ!dZ"dZ#d Z$d!Z%d"Z&d#Z'd$Z(d%Z)e*d&Z+e*d'Z,d(Z-d)Z.d*Z/d+Z0d,Z1d-Z2d.Z3d/Z4d0Z5e*d1Z6d2Z7e*d3Z8e*d4Z9d5Z:d6Z;d7Z<e=e7e9Z>e=e1e4Z?e=e2e5Z@e=e3e6ZAe=e"e)ZBe=e#e+ZCe=e$e,ZDe=e%e-ZEe=e&e.ZFe=e'e/ZGe=e(e0ZHe=e e!ZIe=e:ZJe=e;e<ZKRS(9cCsttj|||_|r(||_nHt|_d|j_d|j_d|j_d|j_d|j_ dS(Nii( tobjectR t _bam_headert_readRtflagtrnametmrnmtpostmpos(R R5t aligned_read((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyR Ws        cCs|j|jdS(s?The header in samfile should match the _bam_header of this readN(twriteR=(R tsamfile((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytwrite_to_samfiledsc Cs|jt|j|jt|jdt|j|j|jt|jdt|j |j |j g }|j }|r|j |j ndj|S(Nis (tqnameR#R>R?RAtmapqtcigarR@RBtisizetseqtqualttagstextendR!(R tstr_listRM((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyR(hs n cCswt|j|jt|jt|jt|jt|j|j|jt|j t|j t j @S(N( thashRGR>R?RARIR@RBRJRKRLtsystmaxint(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt__hash__osc Csy|j|jko|j|jko|j|jko|j|jko|j|jko|j|jko|j|jko|j|jko|j|jko|j |j kSWnt SXdS(N( RGR>R?RARIR@RBRJRKRLtFalse(R tother((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt__eq__rscCs |jjS(N(R=t is_paired(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyRWxscCst|j_dS(N(tTrueR=RW(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_paired_flag{scCs |jjS(N(R=tis_proper_pair(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytis_properly_paired~scCst|j_dS(N(RXR=RZ(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_proper_pair_flagscCs|jjtkS(N(R=t is_unmappedRT(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt is_mappedscCst|j_dS(N(RXR=R](R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_unmapped_flagscCst|j_dS(N(RTR=R](R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_mapped_flagscCst|j_dS(N(RXR=t is_reverse(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_reverse_strand_flagscCs |jjS(N(R=Ra(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt is_reversedscCst|j_dS(N(RTR=Ra(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_forward_strand_flagscCs |jjS(N(R=tmate_is_reverse(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytis_mate_reversedscCst|j_dS(N(RXR=Re(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_mate_reverse_strand_flagscCst|j_dS(N(RTR=Re(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_mate_forward_strand_flagscCs |jjS(N(R=tis_read1(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytis_first_of_pairscCs(t|j_t|j_t|j_dS(N(RXR=RiRTtis_read2RW(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_first_of_pair_flags  cCs |jjS(N(R=Rk(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytis_second_of_pairscCs(t|j_t|j_t|j_dS(N(RTR=RiRXRkRW(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_second_of_pair_flags  cCs%|jjd}tj||S(Ns (R)R*R:t fromFields(R,R5R-((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt fromStringscCst|}|d|_t|d|_|d|_t|dd|_t|d|_|d|_|d|_t|dd|_ t|d |_ |d |_ |d |_ |d |_ |S( Niiiiiiiiii i i (R:RGRR>R?RARHRIR@RBRJRKRLRM(R-R5R+((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyRos        c Csdtj|j|j|j|jd|j|j|j|j d|j |j |j g |j |jS(Ni(R:RoRGR>R?RARHRIR@RBRJRKRLRMR<(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytclonescCs |jjS(N(R=taend(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_end_positionscCs"|j|}|jd|dS(NRR(RR<(R trefnameR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt get_ref_lenscCs.|j|}t||jd|d(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_flagscCs |jjS(N(R=RA(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_posscCs |jjS(N(R=RB(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_mposscCs |jjS(N(R=RJ(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt get_isizescCs|jjr|jjSdS(Nt*(R=RK(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_seqs  cCs|jjr|jjSdS(NR~(R=RL(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_quals  cCs |jjS(N(R=RH(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_mapqscCs||j_dS(N(R=R>(R R((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_flagscCsTt|}|rD||j|jkrD|j|j|dn||j_dS(Ni(RRuR?RwR=RA(R Rt extend_reftval((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_poss cCsTt|}|rD||j|jkrD|j|j|dn||j_dS(Ni(RRuR@RwR=RB(R RRR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_mposs cCs||j_dS(N(R=RJ(R R((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt set_isizescCs||j_dS(N(R=RK(R R((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_seqscCs||j_dS(N(R=RL(R R((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_qual scCs||j_dS(N(R=RH(R R((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_mapq sc Csg}|jj}|rx|D]\}}t|trT|jd||fqt|tr}|jd||fq|jrt|dkr|dkr|jd|tt|fqt|dkr|jd||fq|jd||fqWn|S(Ns%s:i:%ds%s:f:%fitMDs%s:A:%sis%s:Z:%s( R=RMt isinstanceRRtfloattisdigittchrtlen(R ttagstrst read_tagsttagnamettagvalue((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_tagss *&cCspd}g}|jj}|rYx8|jjD]'\}}|jd|||fq+Wn|rldj|SdS(NtMIDNSHPs%d%stR~(R=RIRR!(R t cigar_opcodesR+RItcodetcount((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt get_cigar s " cCs,|jjdkrdS|jj|jjS(NiR~(R=R@R<R (R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_mrnm+scCsg}x|D]}|jdd\}}}|dkrV|j|t|fq |dkr~|j|t|fq |dkr|j||fq |j||fq W||j_dS(Nt:iRtftA(R*RRRR=RM(R ttag_listttuplesttagRttagtypeR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_tags0s    cCs~d}g}|}xY|rmtj|rtj|d\}}}|j|j|t|gqqW||j_dS(NRi(tcigar_opcode_pattsearchR*RtindexRR=RI(R RRR+t remaindertwidth_stopcode((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt set_cigar>s )cCs|j|||j_dS(N(RR=R@(R Rt extend_refs((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytset_mrnmHscCs,|jjdkrdS|jj|jjS(sName of the reference sequence iR~(R=R?R<R (R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt get_rnameKscCs |dkrdS|jj|S(NR~i(R<R(R RR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyRQs cCs|j|||j_dS(N(RR=R?(R RR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt set_rnameVscCs |jjS(sAligned length of the read(R=talen(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytget_alenYscCs|jS(sOThe SAMHeader instance that defines the mapping from rname to tid for this read(R<(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt get_header]scCs ||_dS(N(R<(R R5((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt set_headerasN(LR7R8tNoneR RFR(RSRVRWRYR[R\R^R_R`RbRcRdRfRgRhRjRlRmRnR9RpRoRqRsRuRwRxRyRzR{R|R}RRRRRTRRRRRRRRRRRRRRRRRRtpropertyR?RMRIR@R>RARBRJRKRLRHRGRR5(((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyR:Vs                                                    tSAMReadIteratorcBseZdZdZRS(cCstj|dS(N(R;R (R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyR uscCs tdS(sDummy implementationN(t StopIteration(R ((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytiteratorxs(R7R8R R(((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyRts tSAMReadFileIteratorcBs&eZddZdZdZRS(cCstj||dkr\|jdr1d}q\|jdrId}q\td|ntj|d||_tj |jj |_ dS(Ns.samtrs.bamtrbsH[Reads.samRead.SAMReadFileIterator] Format of SAM/BAM file %s is unknowntmode( RR Rtendswitht ValueErrortpysamtSamfileRERR6R5(R tfilenameR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyR }s    ccs/x(|jD]}t|j|}|Vq WdS(N(RER:R5(R treadtsam_read((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyRsccs/x(|jD]}t|j|}|Vq WdS(N(RER:R5(R RR((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt__iter__sN(R7R8RR RR(((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyR|s cCs}g}t|}xQ|D]I}|jj}|d}|d}d||f}|j|qW|jdj|S(Niis@SQ SN:%s LN:%ss (R/R)R*RR0R!(R1t header_linestfaiR,R-RRthdr_line((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pytgenerate_SAM_header_from_fais     (t__doc__RQtreRRtcompileRRRR;R:RRR(((sQ/woldlab/castor/data00/home/georgi/programs/RNA-Seq-Simulator.v1.0/lib/samRead.pyt s F