o Rŀg v@sdZddlZddlmZddlZddlmZddlmZddl m Z ddl m Z ddl m Z dd l mZGd d d ejZGd d d ejZdS)aBio.Align support for the "sam" pairwise alignment format. The Sequence Alignment/Map (SAM) format, created by Heng Li and Richard Durbin at the Wellcome Trust Sanger Institute, stores a series of alignments to the genome in a single file. Typically they are used for next-generation sequencing data. SAM files store the alignment positions for mapped sequences, and may also store the aligned sequences and other information associated with the sequence. See http://www.htslib.org/ for more information. You are expected to use this module via the Bio.Align functions. Coordinates in the SAM format are defined in terms of one-based start positions; the parser converts these to zero-based coordinates to be consistent with Python and other alignment formats. N)chain) Alignment) interfaces)reverse_complement)Seq)UndefinedSequenceError) SeqRecordcs8eZdZdZdZd fdd ZddZd d d ZZS) AlignmentWriterzGAlignment file writer for the Sequence Alignment/Map (SAM) file format.SAMFcst|||_dS)zCreate an AlignmentWriter object. Arguments: - md - If True, calculate the MD tag from the alignment and include it in the output. If False (default), do not include the MD tag in the output. N)super__init__md)selftargetr  __class__A/var/www/html/myenv/lib/python3.10/site-packages/Bio/Align/sam.pyr +s zAlignmentWriter.__init__c Csz|j}Wn tyi}Ynwz|j}Wn ty!i}Ynw|d}|durVdd|dg}|D]\}}|dkr@q7|d||fq7d|d} || |D]} d g}|d | jt | j } |d | | j D]n\}}|d kr|d |qv|dkr|dd|qv|dkr|d|qv|dkr|d|qv|dkr|d|qv|dkr|dvsJ|d|qv|dkr|d|qv|d|dd|fqvz| j } Wn tyYn w| dkr|d| d|d} || qX|D]6\} }| dkrq|D]'}d| g}|D]\}}|d||fq(d|d} || qqdS) zWrite the SAM header.HDNz@HDzVN:%sVNz%s:%s  z@SQzSN:%szLN:%dalternate_locuszAH:%snameszAN:%s,assemblyzAS:%sMD5zM5:%sspecieszSP:%stopologylinearcircularzPP:%sURIzUR:%szzDS:%s@) metadataAttributeErrortargetsgetitemsappendjoinwriteidlenseq annotations description)rstream alignmentsr%r'valuesfieldskeyvaluelinerecordlengthr1tagrowsrowrrr write_header7sz             zAlignmentWriter.write_headerNc' Cst|ts td|j}|j\}}d}d}z|j}Wn ty)d}d} YnIwz|jd}Wn tt fy<Ynwz|jd}Wn tt fyOYnwz|j d} Wn tt fydd} Yn wd d d | D} |j }t |} z|j} Wn tyd } Ynw|j }|d |d kr|dddddf}|d|dkrd} n d} t|}t|}| |dddf|dddf<||}}z| |jO} Wn tyYnwzt|}WntyYntyd}Ynwt|d}|dddf\}}|d}d}|dur|d|7}|dkr|d|7}z|j}WnUtyxd}|ddddfD]>\}}||}||}|dkrR|d|7}|}q7|dkra|d|7}|}q7||krjtd|d|7}|}|}q7Yn~wt||ddddfD]o\}\}}||}||}|dkr|tdksJ|d|7}|}q|dkr|tdkr|d|7}n|tdkr|d|7}ntd||}q||krtd|td ksJ|d|7}|}|}q|| kr|d| |7}|dur|d|7}z|j}Wn tyd!}Ynwz|j}Wn ty1d}Ynw|| kr9d"}z|j}Wn tyJd}Ynw|d7}d}|t| | t|t|||t|t||| g }|durp|j}|d#ur|dkr~td$|dddf\}}d}d}|dur|ddddfD]j\}}||}||}|dkr|}q|dkr |r|t|7}d}|d%|||7}|}q||krtdt||||||D]\}}||kr|d7}q|t||7}d}q|}|}q|r|t|7}nt||ddddfD]r\}\}}||}||}|dkr4|}q|dkrY|tdkrU|rK|t|7}d}|d%|||7}|}q||krbtdt||||||D]\}}||kr~|d7}qo|t||7}d}qo|}|}q|r|t|7}d&|} || z|j}!Wn tyYn wd'|!} || z|j}"Wn tyYnw|"D]\}#}$t|$t rd(}%t|$}$nat|$t!rd)}%t|$}$nTt|$trt |$dkrd*}%nDd+}%nAt|$trd,}%d t"t|$}$n0t|$tjrAd-}%t#|$j$tj%r$nt#|$j$t!r-n td.|$j$d/|#d0d t"t|$}$|#d1|%d1|$} || qd2 |d3}&|&S)4zBReturn a string with a single alignment formatted as one SAM line.zExpected an Alignment objectNquery*hard_clip_lefthard_clip_right phred_qualitycss|] }t|dVqdS)!N)chr).0r7rrr sz3AlignmentWriter.format_alignment..r)rr)rrI)r)rIrJrrJASCIIz%dHz%dSz%dIz%dDzUnequal step sizes in alignmentz%dMINz%dNDzUnexpected operation M=Tz(requested MD tag with undefined sequence^zMD:Z:%sz AS:i:%.0fifAZHBz Array of incompatible data type z in annotation '':rr)& isinstancer TypeError coordinates transpose sequencesr-r&r0KeyErrorletter_annotationsr+r/r.rnparrayflagbytesrstr operations ValueErrorzipordmapqrnextpnextr r*scorer)intfloatmap issubdtypedtypeinteger)'r alignmentr r^rr?rArBqNamequalphredqSizernameretStartqStartposcigarrhtEndqEndtCountqCount operationrlrmrntLenr5numbertcqcfieldror0r6r7datatyper8rrrformat_alignmentus                          &                    "   &    "             z AlignmentWriter.format_alignment)F)N) __name__ __module__ __qualname____doc__fmtr r>r __classcell__rrrrr &s  >r c@s$eZdZdZdZddZddZdS)AlignmentIteratoraAlignment iterator for Sequence Alignment/Map (SAM) files. Each line in the file contains one genomic alignment, which are loaded and returned incrementally. The following columns are stored as attributes of the alignment: - flag: The FLAG combination of bitwise flags; - mapq: Mapping Quality (only stored if available) - rnext: Reference sequence name of the primary alignment of the next read in the alignment (only stored if available) - pnext: Zero-based position of the primary alignment of the next read in the template (only stored if available) - tlen: signed observed template length (only stored if available) Other information associated with the alignment by its tags are stored in the annotations attribute of each alignment. Any hard clipping (clipped sequences not present in the query sequence) are stored as 'hard_clip_left' and 'hard_clip_right' in the annotations dictionary attribute of the query sequence record. The sequence quality, if available, is stored as 'phred_quality' in the letter_annotations dictionary attribute of the query sequence record. r cCsi|_g|_|D]}|ds||_n|ddd}|d}i}|dkri}d}|ddD]u}|dd\} } t| dksEJ| dkrL| } q3| d krUt| } q3| d kr^| |d <q3| d krj| d |d<q3| dkrs| |d<q3| dkrz| }q3| dkr| |d<q3| dkr| |d<q3| dkr| dvsJ| |d<q3| dkr| |d<q3| || <q3td| d} t | | d|d}|dur||_ |j |q|ddD]}|dd\} } t| dksJ| || <q|dkr||j|<q||jvrg|j|<|j| |qdd t |jD|_ dS)!Nr$rJrrSQr[r#SNLNAHrANrrASrDSM5rSPrTPrrURr"r:rD)r-r1r0rcSsi|]\}}|j|qSr)r-)rGindexr9rrr s z2AlignmentIterator._read_header..)r%r' startswith_linestripsplitr.rprrr1r* enumerate_target_indices)rr2r8r5r;r4r0r1rr6r7r{r:sequencer9rrr _read_headersp                zAlignmentIterator._read_headerc/sZ z|j}Wn ty|}Yn wt|g|}|`|D]}|}t|dkr0tdt||d}t|d}|d}t|dd}t|d} |d} |d } t|d d} t|d } |d }|d }d}d}i}|ddD]}|dd\}}|dkr|dksJt}qv|dkr|dksJ}qv|dkrtnV|dkrtnM|dvrnH|dkrt}tfddt d|dDn0|dkrd}ddd|dvrt}n|dkrt}n td|d|dt |||<qv|d@r d}nd }d}d}d!}|d@rd}d}n|durd}||gg}d"} t }!| D]}|d#krCt| }"||"7}||"7}n|d$vrWt| }"||"7}||"7}d%}nr|d&kret| }"||"7}nd|d'krt| }"|dkr}|dd|"7<||"7}d"} q/|d(krt| }"||"7}n6|d)krt| }"||"7}d%}n&|dkr|dkrt| }nt| }d"} q/|d*krt d+| |7} q/|||g|!t|d"} q/|j|}#|#dur|jrtd,|d-td|d"d.}n|j|#}nd}||gg}|}$d"}|g}%d}&g}'d"} t }!| D]}|d#vrAt| }"||"7}||"7}||$d|"7}|$|"d}$|&|"7}&n|d$vrgt| }"||"7}||"7}||$d|"7}|$|"d}$|&|"7}&d%}n|d&kr{t| }"||"7}|$|"d}$n|d'krt| }"|dkr|dd|"7<||"7}|$|"d}$d"} q|d(krt| }"||"7}|&|"7}&|%||'|&d}&nB|d)krt| }"||"7}|%||'|&d}&d%}n&|dkr|dkrt| }nt| }d"} q|d*krt d+| |7} q|||g|!t|d"} q|'|&|}$d"}d"} t|}(|(D]b}|d/vrN| rBt| } ||$d| 7}|$| d}$d"} ||7}|$dd}$q$|d0kr| rjt| } ||$d| 7}|$| d}$d"} |(D]}|d/vrun||7}qln |} q$| |7} q$| rt| } ||$d| 7}|}$|j|}#t|j|#}t|j}"i})d}#t|%|'D]\}*}&|$|#|#|&|)|*<|#|&7}#qt|)|"d1|_|durt |}|dkr||dddf|dddf<|d2kr|}"td|"d1}+nt|}+|d@st||ksJ|dkr|+}+t|+|d"d.}|dkr)||}}|dur3||jd3<|dur=||jd4<|d2krNd5d6|D},|,|jd7<||g}-t|-|}.||._| d8krb| |._ | d9krk||._!n| d2krs| |._!| dkr{| |._"| dkr| |._#|dur||._$|r||._|dur||._%|dur||._&|r|!|._'|.SdS):N z)line has %d columns; expected at least 11rrJr# r[rrTMDrWrU)rVrWrXc3s$|] }t||dVqdS)r#N)rp)rGrTr7rrrHs"z9AlignmentIterator._read_next_alignment..rYrcCsSiIzUnknown number type 'z ' in tag 'rZrK-+FrDrPz=XTrMSrOrNPz'padding operator is not yet implementedz Found target z missing from header)r-r1 ACGTNacgtnrSrr@rArBcSsg|]}t|dqS)rE)rk)rGcrrr sz:AlignmentIterator._read_next_alignment..rCrQrR)(rr&rrr.rirprqrfrangercrd bytearrayNotImplementedErrorr*rkrr(r'ritercopydeepcopyr/rjrr_rr0rbrrerlrmrntlenrorArBrh)/rr2r8linesr5qnamerer{ target_posrlrrmrnrr?rxr ror0rr;rnletterrtstrandrArBstore_operationsrr^ query_posrrhr:rr/startssizesizeslettersdatastartrryrecordsrvrrr_read_next_alignments"                                                                                           z&AlignmentIterator._read_next_alignmentN)rrrrrrrrrrrrts  =r)rr itertoolsrnumpyrc Bio.AlignrrBio.Seqrrr Bio.SeqRecordrr rrrrrs       P