:>"^8ã@spdZddlmZddlmZddlmZddlmZddl m Z Gdd „d e ƒZ d S) ajBio.AlignIO support for GCG MSF format. The file format was produced by the GCG PileUp and and LocalPileUp tools, and later tools such as T-COFFEE and MUSCLE support it as an optional output format. The original GCG toola would write gaps at ends of each sequence which could be missing data as tildes (``~``), whereas internal gaps were periods (``.``) instead. This parser replaces both with minus signs (``-``) for consistency with the rest of ``Bio.AlignIO``. You are expected to use this module via the Bio.AlignIO functions (or the Bio.SeqIO functions if you want to work directly with the gapped sequences). é)Úprint_function)ÚSeq)Ú SeqRecord)ÚMultipleSeqAlignmenté)ÚAlignmentIteratorc@s(eZdZdZdZdd„ZdS)Ú MsfIteratorzGCG MSF alignment iterator.Ncsœˆj}ˆjdkr'|jƒ}nˆj}dˆ_|sEt‚dddg}|jƒjƒd|krŸtd|jƒjƒddj|ƒfƒ‚x"|rÃd|krÃ|jƒ}q¢W|sÖtd ƒ‚|jd ƒjƒ}|jd ƒ}||d d ks.|d<d=ks.|d>dkr>td|ƒ‚yt ||dƒ}Wntk rpd?}YnX|dkr•td||dƒ‚||d}|d@kr¿td|ƒ‚g}g} g} g} |jƒ}x‘|rv|jƒdkrv|jƒ}|jƒj dƒræd|krcd|krcd|krc||jdƒdd…jƒ} | jdƒ\} } | jdƒ\}} | jdƒ\}}| jƒ} | j dƒrÐ| ddA…} | |krìtd| ƒ‚d | krt d!| ƒ‚|j | ƒ| j t |jƒƒƒ| j t |jƒƒƒ| j t|jƒƒƒqætd"|ƒ‚qæW|s‰td#ƒ‚|t| ƒkrèt| ƒ‰t‡fd$d%†| Dƒƒ}td&||t|ƒˆfƒ‚|jƒ}|std'ƒ‚|jƒrtd(ƒ‚d)d*„|Dƒ}d}x×||krx’t|ƒD]„\}} |jƒ}|dkrª|jƒ rªx#|r©|jƒ r©|jƒ}q‡W|s¼td+ƒ‚|jƒjƒ}|dkr |r |d| kr yt |dƒ}Wntk rdB}YnX||dkrItd,|d|fƒ‚t|ƒdkrt|ƒd krvdC}n/yt |dƒ}Wntk r¤dD}YnX||d-|krÂ|d-n|krtd.|d|d-|krò|d-n||fƒ‚|jƒ}|jƒjƒ}|su| ||kr\td/j||ƒƒ| |kr\q×td0| |fƒ‚qS|d| krÁt|ƒdks£t|ƒ‚||j|dd…ƒqStd1| |fƒ‚qSW|d-7}|jƒ}|jƒr:td2|ƒ‚q:Wxb|jƒ}|s*Pq|jƒs9q|jƒjƒd|krb|ˆ_Pqtd3|ƒ‚qWd4d*„|Dƒ}d5}xott| |ƒƒD]X\}\}}t|ƒ|kr¥t|ƒ|kr¥d6}|d7|t|ƒ|| MSF: Type: Check: ..', not: %rzCGCG MSF header line should have MDF: for column count, not %rÚPÚNz]GCG MSF header line should have 'Type: P' (protein) or 'Type: N' (nucleotide), not 'Type: %s'z//zName: z Len: z Check: z Weight: éz oozDuplicated ID of %rú zSpace in ID %rzMalformed GCG MSF name line: %rz4End of file while looking for end of header // line.c3s!|]}|ˆkrdVqdS)rN©)Ú.0Ú_)Ú max_lengthrú6/tmp/pip-build-ww9dw3qa/biopython/Bio/AlignIO/MsfIO.pyú Àsz'MsfIterator.__next__..zLGCG MSF header said alignment length %i, but %s of %i sequences said Len: %sz.End of file after // line, expected sequences.z4After // line, expected blank line before sequences.cSsg|] }g‘qSrr)rrrrrú Ís z(MsfIterator.__next__..z*End of file where expecting sequence data.z5Expected GCG MSF coordinate line starting %i, got: %ré2z2Expected GCG MSF coordinate line %i to %i, got: %rÚz!Expected sequence for %s, got: %rz!Expected sequence for %r, got: %rzExpected blank line, got: %rz+Unexpected line after GCG MSF alignment: %rcSs7g|]-}dj|ƒjddƒjddƒ‘qS)rú~ú-Ú.)ÚjoinÚreplace)rÚsrrrr/s FTr)ÚBiopythonParserWarningzGOne of more alignment sequences were truncated and have been gap paddedc 3sN|]D\}}}tt|ˆjƒd|d|d|dd|iƒVqdS)ÚidÚnameÚ descriptionÚ annotationsÚweightN)rrÚalphabet)rÚir Úw)ÚselfrrrAsz5GCG MSF headers said alignment length %i, but have %iéýÿÿÿ)r r éÿÿÿÿr,)rrr+r,r,r,)ÚhandleÚ_headerÚreadlineÚ StopIterationÚstripÚsplitÚ ValueErrorrÚindexÚintÚ startswithÚendswithÚNotImplementedErrorÚappendÚfloatÚmaxÚsumÚlenÚ enumerateÚAssertionErrorÚextendÚzipÚwarningsZBior!Úwarnrr'Zget_alignment_length)r*r-ÚlineZ known_headersÚpartsÚoffsetZ aln_lengthZseq_typeÚidsZlengthsZchecksZweightsÚrestr#ÚlengthÚcheckr&Ú max_countZseqsZcompleted_lengthÚidxÚwordsr(Zpaddedr rBr!ÚrecordsZalignr)rr*rÚ__next__$s:   ) '        $#            "        #      ($  zMsfIterator.__next__)Ú__name__Ú __module__Ú __qualname__Ú__doc__r.rOrrrrrs rN) rSÚ __future__rZBio.SeqrZ Bio.SeqRecordrZ Bio.AlignrZ InterfacesrrrrrrÚs