o Rŀg7@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}|jdur |}n|j}d|_|stgd}|d|vr6td|dd|f|rFd|vrF|}|rFd|vs<|sLtd|d}|d }||d d ksl|d d vsl|ddkrrtd|z t ||d}Wn tyd}Ynw|dkrtd||d||d}|dvrtd|g}g} g} g} |}|rR|dkrR|}| drId|vrBd|vrBd|vrB||ddd} | d\} } | d\}} | d\}}| } | dr | dd } | |vrtd| d| vr!t d| | | | t || t || t|ntd ||rR|dks|sYtd!|t| kr{t| tfd"d#| D}td$||t|f|}|std%|rtd&d'd(|D}d}||krt|D] \}} |}|dkr|s|r|s|}|r|r|std)|}|dkr`|r`|d| kr`zt |d}Wn tyd}Ynw||dkr td*|d|ft|dkrVt|d krd}nzt |d}Wn ty/d}Ynw||d+|kr<|d+n|krVtd,|d|d+|krP|d+n||f|}|}|s| ||krztd-||| |krzqtd.| d/||d| krt|dksJ||||ddqtd.| d/||d+7}|}|rtd0|||ks |}|sn|sn|d|vr||_n td2|qd3d(|D}d4}tt| |D]#\}\}}t||krt||krd1}|d5|t|||<q|r4ddl}dd6lm}|d7|d8d#t||| D}t|}||krTtd9||f|S):z)Parse the next alignment from the handle.N)z!!NA_MULTIPLE_ALIGNMENTz!!AA_MULTIPLE_ALIGNMENTPileUprz$%s is not a known GCG MSF header: %sz, z MSF: z6Reached end of file without MSF/Type/Check header line zMSF:zType:)zCheck:z CompCheck:z..zsGCG MSF header line should be ' 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  z Space in ID zMalformed GCG MSF name line: z4End of file while looking for end of header // line.c3s|] }|krdVqdS)rN.0_ max_lengthrE/var/www/html/myenv/lib/python3.10/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|]}gqSrrrrrr 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: %rzExpected sequence for z, got: zExpected blank line, got: Tz)Unexpected line after GCG MSF alignment: cSs&g|]}d|ddddqS)r~-.)joinreplace)rsrrrr.s&Fr)BiopythonParserWarningzGOne of more alignment sequences were truncated and have been gap paddedcss0|]\}}}tt||||d|idVqdS)weight)idname description annotationsN)rr)rir"wrrrr@s  z5GCG MSF headers said alignment length %i, but have %i)handle_headerreadline StopIterationstripsplit ValueErrorr indexint startswithendswithNotImplementedErrorappendfloatmaxsumlen enumerateextendzipwarningsBior#warnrget_alignment_length)selfr+line known_headerspartsoffset aln_lengthseq_typeidslengthschecksweightsrestr&lengthcheckr$ max_countseqscompleted_lengthidxwordsr)paddedr"r?r#recordsalignrrr__next__#sv    '                  P    zMsfIterator.__next__)__name__ __module__ __qualname____doc__r,rYrrrrrs rN) r] Bio.AlignrBio.Seqr Bio.SeqRecordr Interfacesrrrrrrs