o RŀgF@sddZddlZddlmZddlmZddlmZddlmZGdddej Z Gd d d ej Z dS) axBio.Align support for the "maf" multiple alignment format. The Multiple Alignment Format, described by UCSC, stores a series of multiple alignments in a single file. It is suitable for whole-genome to whole-genome alignments, metadata such as source chromosome, start position, size, and strand can be stored. See http://genome.ucsc.edu/FAQ/FAQformat.html#format5 You are expected to use this module via the Bio.Align functions. Coordinates in the MAF format are defined in terms of zero-based start positions (like Python) and aligning region sizes. A minimal aligned region of length one and starting at first position in the source sequence would have ``start == 0`` and ``size == 1``. As we can see on this example, ``start + size`` will give one more than the zero-based end position. We can therefore manipulate ``start`` and ``start + size`` as python list slice boundaries. N) Alignment) interfaces)Seq) SeqRecordc@s4eZdZdZdZddZddZddZd d Zd S) AlignmentWriterz-Accepts Alignment objects, writes a MAF file.MAFcCs|d|D]E\}}|dvrn)|dkr!|dvr td|n|dkr0|dvr/td|n |d kr:d |}nq d |vrCd |}|d |d |q |d dS)Ntrackname descriptionframesmafDotonoffz2mafDot value must be 'on' or 'off' (received '%s') visibilitydensepackfullzCvisibility value must be 'dense', 'pack', or 'full' (received '%s') speciesOrder z"%s"= )writeitems ValueErrorjoin)selfstreammetadatakeyvaluer#A/var/www/html/myenv/lib/python3.10/site-packages/Bio/Align/maf.py_write_trackline-s2  z AlignmentWriter._write_tracklinec Cs"z|j}Wn tyddi}Ynwd}|D]}||vr%|||nq|d|D]<\}}||vr8q/|dkr=q/|dkrL|dkrItdd}n|dkrSd }n |d krZd }ntd ||d |d|q/|d|d}|dur|D] }|d|dq||ddSdS)zWrite the MAF header. MAF Version1)r r r r rr##mafCommentsMAF version must be 1versionScoringscoringProgramprogramzUnexpected key '%s' for headerrrrNz# )r AttributeErrorr%rrrget) rr alignmentsr track_keysr!r"commentscommentr#r#r$ write_headerFsD        zAlignmentWriter.write_headercCsVz|j}Wn tyd}Ynwd|d}|d}|dur'|d|7}|dS)Naza score=z.6fpassz pass=r)scorer0r1)r alignment annotationsr9liner"r#r#r$_format_score_lineps    z"AlignmentWriter._format_score_linec"Cs0t|ts tdz|j}Wn tyi}Ynwg}|||}||d}d}d}d}t|j} t | D]^} |j| } |j | } z| j } Wn tyVd| } Ynw| d}| d}t| }||krl||}n||}||}t |t| }t |tt |}t |tt |}t |tt |}q8t|dgD]Z\} }|\} }}z| j } Wntyd| | } Ynwt |t| }|\}}t| j}||kr||}n||}||}t |tt |}t |tt |}t |tt |}q||||d}t | D]} |j| } |j | } z| j }Wnty(d| }Ynw| d}| d}t| }||krA||}d}n ||}||}d}|| }||} t ||}t ||}t ||}d | d |d |d |d |d |d }||z| j}Wn tyd }Ynw|d ur|d }|d urd}d}|D]}|dkr|d7}q|||7}|d7}q||} d| d |d }||z|d}|d}|d} |d}!Wn tyYqw||} d| d |d |d | d |!d }||qt|dgD]s\} }|\} }}z| j } Wnty8d| | } Ynw| |} |\}}t| j}||krS||}d}n ||}||}d}t ||}t ||}t ||}d| d |d |d |d |d |d }||q|d d|S)zAReturn a string with a single alignment formatted as a MAF block.zExpected an Alignment objectrz sequence_%dempty+-s rrNqualityq leftStatus leftCount rightStatus rightCounti ze ) isinstancer TypeErrorr;r0r=appendlen sequencesrange coordinatesidmaxstr enumerater1seqljustrjustKeyErrorr)"rr:alignment_annotationslinesr< name_width start_width size_width length_widthnirecordrSr startendlengthsizer?segmentstatus quality_width record_idstrandtextr;rDgapped_qualityjletterrHrIrJrKr#r#r$format_alignment|s                        *             $      *  z AlignmentWriter.format_alignmentN) __name__ __module__ __qualname____doc__fmtr%r6r=rrr#r#r#r$r(s* rc@s,eZdZdZdZdZdZddZddZd S) AlignmentIteratoraJAlignment iterator for Multiple Alignment Format files. The file may contain multiple concatenated alignments, which are loaded and returned incrementally. File meta-data are stored in the ``.metadata`` attribute of the returned iterator. Alignment annotations are stored in the ``.annotations`` attribute of the ``Alignment`` object, except for the alignment score, which is stored as an attribute. Sequence information of empty parts in the alignment block (sequences that connect the previous alignment block to the next alignment block, but do not align to the current alignment block) is stored in the alignment annotations under the ``"empty"`` key. Annotations specific to each line in the alignment are stored in the ``.annotations`` attribute of the corresponding sequence record. r)CINrbMT)ryrzr|rbc Csi}t|}|drZt|}|ddD]?}|d\}}|dvr$n-|dkr3|dvr2td|n|dkrB|d vrAtd |n|d krK|}ntd ||||<qt|}|}|d dkrhtd|ddD](}|d\}}|dkr~d}n|dkrd}n |dkrd}ntd||||<qn|ddkrtdg}|D]$}|r|ds|dsJ||_n|dd} || q| |r||d<||_ dS)Nztrack rFrr r rz7Variable mafDot in track line has unexpected value '%s'rrz;Variable visibility in track line has unexpected value '%s'rz&Unexpected variable '%s' in track linerr(z%header line does not start with ##mafr+r&r-r,r/r.z'Unexpected variable '%s' in header liner'r*#r7r)) next startswithshlexsplitrr1strip_alinerO_closer ) rrr r<wordswordr!r"r4r5r#r#r$ _read_headerst           zAlignmentIterator._read_headerc&Cs|j}|dur dSg}g}g}g}d}g}i} |dd} | D].} | d\} } | dkr3t| }q!| dkrJt| } | dkrEtd| | | d<q!td| |D]}|dr[qR|d rf||_nu|d r|} t| d kr{td | d}t| d }t| d}| d}t| d}| d}dD]}||d}q| | t d|d}t ||ddd}| || || || |qR|dr#|} t| dksJ| d|ksJ| d }t| d}| d}t| d}|t jvsJ|t jvsJ||jd<||jd<||jd<||jd<qR|dr|dd} t| dks:J| d}t| d}t| d }| d}t| d}| d}|t jvs`Jt d|d}t ||ddd}||}|dkr|||f}n||||f}|||f} | d}!|!durg}!|!| d<|! | qR|dr|} t| dksJ| d|ksJ| d dd} | |jd <qR|sqRtd!|d|_t|\}"}#t|||"|D]&\}}}}t|j}t||krtd"t||ft ||i|d|_qt|||#D]+\}}}$|dkr4|$d#|$d|$|$dd<|j|_|jjdd}|$|7}$qt||#}%| durO| |%_|durW||%_|%S)$NrFrr9r8rz&pass value must be positive (found %d)z Unknown annotation variable '%s'r~r7rCz5Error parsing alignment - 's' line must have 7 fieldsr@z.=_rB)rgrE)rTr r rLrHrIrJrKerAr?rGrDz+Error parsing alignment - unexpected line: z2sequence size is incorrect (found %d, expected %d)r>)rrfloatintrrrrPreplacerOencoderrrxstatus_charactersr;empty_status_charactersr1rparse_printed_alignmentziprXreverse_complementdefined_rangesr9)&rralinerecordsstartssizesstrandsr9aligned_sequencesr;rrr!r"r<srcrerhrmsrcSizerngap_charrXrdrHrIrJrKrjsequencerfrir? annotationrQrSrowr:r#r#r$_read_next_alignmentOs                                                 z&AlignmentIterator._read_next_alignmentN) rsrtrurvrwrrrrr#r#r#r$rxs :rx) rvr Bio.AlignrrBio.Seqr Bio.SeqRecordrrrxr#r#r#r$s     X