o Rŀg5 @sdZddlZddlZddlZddlZddlmZddlmZddlm Z ddlm Z ddlm Z ddl m Z dd l mZdd lmZdd lmZe d d eddddeddddeddddeddddgZGddde jZGddde je jZdS)a'Bio.Align support for the "bigmaf" multiple alignment format. The bigMaf format stores multiple alignments in a format compatible with the MAF (Multiple Alignment Format) format. BigMaf files are binary and are indexed as a bigBed file. See https://genome.ucsc.edu/goldenPath/help/bigMaf.html N) _aligncore) Alignment) Alignments)bigbed)maf) AutoSQLTable)Field)Seq) SeqRecordbedMafzBed3 with MAF blockstringchromz)Reference sequence chromosome or scaffold)as_typenamecommentuint chromStartzStart position in chromosomechromEndzEnd position in chromosomelstringmafBlockz MAF blockcs6eZdZdZdZ    d fdd Zd d ZZS) AlignmentWriterz1Alignment file writer for the bigMaf file format.bigMafNTc stj|dt||||ddS)awCreate an AlignmentWriter object. Arguments: - target - output stream or file name. - targets - A list of SeqRecord objects with the chromosomes in the order as they appear in the alignments. The sequence contents in each SeqRecord may be undefined, but the sequence length must be defined, as in this example: SeqRecord(Seq(None, length=248956422), id="chr1") If targets is None (the default value), the alignments must have an attribute .targets providing the list of SeqRecord objects. - compress - If True (default), compress data using zlib. If False, do not compress data. Use compress=False for faster searching. - blockSize - Number of items to bundle in r-tree. See UCSC's bedToBigBed program for more information. Default value is 256. - itemsPerSlot - Number of data points bundled at lowest level. See UCSC's bedToBigBed program for more information. Use itemsPerSlot=1 for faster searching. Default value is 512. )bedN declarationtargetscompress blockSize itemsPerSlotN)super__init__r)selftargetrrrr  __class__D/var/www/html/myenv/lib/python3.10/site-packages/Bio/Align/bigmaf.pyr"As# zAlignmentWriter.__init__c Cs t}|D]J}t|tstdt|ddddd}|j}|js$q|dd}|jj dd \}}||j_ |d |d ksBJi|_ ||j d <| |q|j d dd|jd} | j dd \}}t|j} t| j|d| d<| |_tj|dt|jd|dS)zWrite the file.zExpected an Alignment objectrN ;.rr)rr)rcSs|jj|jdfS)Nr/)r$id coordinates) alignmentr'r'r(sz,AlignmentWriter.write_file..)keyr)r0r)rrr)r isinstancer TypeErrorformatreplacer1sizer$r0split annotationsappendsortrlistr seqrrrrwrite) r#stream alignmentsfixed_alignmentsr2rr1 reference chromosomerecordrr'r'r( write_filens6        zAlignmentWriter.write_file)NTrr)__name__ __module__ __qualname____doc__fmtr"rG __classcell__r'r'r%r(r<s-rcsDeZdZdZdZdZfddZddZfdd Zd d Z Z S) AlignmentIteratoraAlignment iterator for bigMaf files. The file may contain multiple alignments, which are loaded and returned incrementally. Alignment annotations are stored in the ``.annotations`` attribute of the ``Alignment`` object, except for the alignment score, which is stored as an attribute. Sequence information of empty parts in the alignment block (sequences that connect the previous alignment block to the next alignment block, but do not align to the current alignment block) is stored in the alignment annotations under the ``"empty"`` key. Annotations specific to each line in the alignment are stored in the ``.annotations`` attribute of the corresponding sequence record. rbcsd|_t|dS)z|Create an AlignmentIterator object. Arguments: - source - input file stream, or path to input file N)rDr!r")r#sourcer%r'r(r"szAlignmentIterator.__init__cCst|jd}|j}|j} z|j}Wn tyYnw|d}q|}||j d}d}d} | |} |j rJ| | 7} t } | | }n|| 7}z|d|} Wn ty_Yq3w|| dd} t| dkrqnq4| d}|||d d\}}|S) NIIITrrs;sr.r,.)structStruct byteorderr9treechildrenAttributeErrortellseek dataOffsetread _compressedzlib decompressobj decompressindex ValueErrorr:lendecode)r#rA formatterr9noderXfileposdataSizedatacompressed_datachunk decompressoriwordsrrDrEr'r'r(_read_referencesH        z!AlignmentIterator._read_referencecsJt||jdur|||_d|_d|j|jdjf|jd_dS)Nrz%s.%s)r! _read_headerrDrp_indexrr0)r#rAr%r'r(rqs   "zAlignmentIterator._read_headerc-Cs||ddks Jt|}|||}g}g} i} d} td} d} g} | d}|||d}|dkrJtd||d}||d} nJ|dkrtd||d}||d} ||| }|dd}|D]0}|d \}}|d krt|} qr|d krt |}|dkrt d ||| d <qrt d| n|dkr1td||d}||d} ||| }|dd}t |dkrt d|d }t |d}t |d}|d}t |d}| }| |||\}}t ||kr t dt ||ft||i|d}t||ddd}||||} | d}| |nc|dkrtd||d}||d} ||| }|dd}t |dks_J|d |ksjJ|d } t |d}!|d }"t |d}#| tjvsJ|"tjvsJ| |jd<|!|jd<|"|jd<|#|jd<n|d kr>td||d}||d} ||| }|dd}t |d!ksJ|d }t |d}t |d}|d}t |d}|d }$|$tjvsJtd|d}t||ddd}||}%|d"kr||%f}&n||||%f}&||&|$f}'| d#}(|(dur8g}(|(| d#<|(|'nV|d$krtd||d}||d} ||| }|dd}t |dkslJ|d |kswJ|dd%d&}| |jd'<n |d(krnt d)|q&| j})t|)tj}*| |*t|| |*D]+\}}}+|d"kr|+|jjdd7}+q|j|_|jjdd|+|+dd<qt ||*},| dur| |,_| dur| |,_!|,S)*Nr.r;r)T#s^[^;]*a=sscorespassz&pass value must be positive (found %d)passz Unknown annotation variable '%s'ss$^s\s*\S*\s*\d*\s*\d*\s*[+-]\s*\d*\s*z5Error parsing alignment - 's' line must have 7 fieldsr,rz2sequence size is incorrect (found %d, expected %d))length)r0r descriptioni leftStatus leftCount rightStatus rightCounte+emptyq-rRqualityz+Error parsing alignment - unexpected line: )" memoryviewrPrintedAlignmentParserrematchspantobytesr:floatintrcrerdfeedr r r<rNstatus_charactersr;empty_status_charactersgetr8shapenprint64fillzipr?defined_rangesreverse_complementrscore)-r#chromIdrrrj dataStartdataEndbufferrecordsstrandsr;rprinted_alignment_parserj sequencesrnprefixmlinerowordr4valuesrcstartr9strandsrcSizensequencer?rFrrrrstatusendsegmentr annotationrr1rowr2r'r'r(_create_alignments                                          n       z#AlignmentIterator._create_alignment) rHrIrJrKrLmoder"rprqrrMr'r'r%r(rNs  )rN)rKrrTr_numpyr Bio.AlignrrrrrBio.Align.bigbedrrBio.Seqr Bio.SeqRecordr rrrNr'r'r'r(sR          P