B bd7@sHdZddlmZddlmZddlmZddlmZGdddeZ d S) aiBio.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 tool 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). )MultipleSeqAlignment)Seq) SeqRecord)AlignmentIteratorc@seZdZdZdZddZdS) MsfIteratorzGCG MSF alignment iterator.Nc s|j}|jdkr|}n |j}d|_|s.tdddg}|d|krntd|dd|fx|rd|kr|}qpW|std |d }|d }||d d ks|ddks|ddkrtd|yt ||d}Wntk rd}YnX|dkr.td||d||d}|dkrPtd|g}g} g} g} |}x.|r|dkr|}| drld|krd|krd|kr||ddd} | d\} } | d\}} | d\}}| } | dr| dd} | |kr0td| d | krFt d!| | | | t || t || t|n td"|qlW|std#|t| krt| tfd$d%| D}td&||t|f|}|std'|rtd(d)d*|D}d}xR||krzxt|D] \}} |}|dkr|sx|r|s|}qhW|std+|}|dkr|r|d| kryt |d}Wntk rd}YnX||dkr td,|d|ft|dkrt|d kr.d}n,yt |d}Wntk rXd}YnX||d-|krr|d-n|krtd.|d|d-|kr|d-n||f|}|}|s| ||krtd/||| |krntd0| |fnL|d| kr:t|dks"t||||ddntd1| |fq@W|d-7}|}|r*td2|q*WxP|}|sPn8|sn,|d|kr||_Pn td3|q~Wd4d*|D}d5}xXtt| |D]F\}\}}t||krt||krd6}|d7|t|||<qW|r`ddl}dd8lm}|d9|d:d%t||| D}t|}||krtd;||f|S) MSF: Type: Check: ..', not: %rrzCGCG MSF header line should have MDF: for column count, not %r)PNz]GCG MSF header line should have 'Type: P' (protein) or 'Type: N' (nucleotide), not 'Type: %s'z//zName: z Len: z Check: z Weight: z oozDuplicated ID of %r zSpace in ID %rzMalformed GCG MSF name line: %rz4End of file while looking for end of header // line.c3s|]}|krdVqdS)rN).0_) max_lengthr0lib/python3.7/site-packages/Bio/AlignIO/MsfIO.py sz'MsfIterator.__next__..zLGCG MSF header said alignment length %i, but %s of %i sequences said Len: %sz.End of file after // line, expected sequences.z4After // line, expected blank line before sequences.cSsg|]}gqSrr)rrrrr sz(MsfIterator.__next__..z*End of file where expecting sequence data.z5Expected GCG MSF coordinate line starting %i, got: %r2z2Expected GCG MSF coordinate line %i to %i, got: %rz!Expected sequence for %s, got: %rz!Expected sequence for %r, got: %rzExpected blank line, got: %rz+Unexpected line after GCG MSF alignment: %rcSs&g|]}d|ddddqS)r~-.)joinreplace)rsrrrr-sFTr)BiopythonParserWarningzGOne of more alignment sequences were truncated and have been gap paddedcss.|]&\}}}tt||||d|idVqdS)weight)idnameZ descriptionZ annotationsN)rr)rirwrrrr?sz5GCG MSF headers said alignment length %i, but have %i)handle_headerreadline StopIterationstripsplit ValueErrorrindexint startswithendswithNotImplementedErrorappendfloatmaxsumlen enumerateAssertionErrorextendzipwarningsZBior warnrZget_alignment_length)selfr&lineZ known_headerspartsoffsetZ aln_lengthZseq_typeZidsZlengthsZchecksZweightsrestr#lengthZcheckr!Z max_countZseqsZcompleted_lengthidxZwordsr$Zpaddedrr;r ZrecordsZalignr)rr__next__"s:    '                       zMsfIterator.__next__)__name__ __module__ __qualname____doc__r'rDrrrrrsrN) rHZ Bio.AlignrZBio.SeqrZ Bio.SeqRecordrZ Interfacesrrrrrrs