Ri %vdZddlZddlmZddlmZddlmZddlmZddl m Z Gdd ejZ y) aBio.Align support for GCG MSF format. The file format was produced by the GCG PileUp and LocalPileUp tools, and later tools such as T-COFFEE and MUSCLE support it as an optional output format. You are expected to use this module via the Bio.Align functions. N)BiopythonParserWarning) Alignment) interfaces)Seq) SeqRecordceZdZdZdZdZy)AlignmentIteratorzGCG MSF alignment iterator.MSFc r |j}|s |jdk(r tddygd}|jj d|vr>t|jj dddj ||D]+}|j d}d|vs|jds+n td |j }|jd }||d zd k7s|d dvrtd|z t||dz}||dz}|dvrtd|zg}g} g} g} |D] }|j}|dk(rn|jds-|j } | jd} | jd}| jd}| jd}| | dz}t| |dz}t| |dz}| |dz}||vrtd||j|| j|| j|| j| td|j}|s tdd|jr tdd gt|z}|D]}|j } | s| d} |j|}d j | dd}| |t||jd#z z }|dkrtd$|z||xx|z cc<|| |<t!d%| Dsn td&|D]}|jd k(rJt#d'|D}||k7r#t%j&d(||fzt(d )|}t+|D]]\}}d j |j-d*d#j-d+d#}t||kr|d#|t|z zz }|||<_|Dcgc]}|j/}}t1j2|\}}t5||| Dcgc]!\}}}t7t9||||d,|i-#}}}}t1||}|j:}||k7rtd(||fz|S#t$rtd||dzzdwxYw#t$rtd|dwxYw#t$r-| D]}|jrnYmtd!|d"dwxYwcc}wcc}}}w).Nrz Empty file.)z!!NA_MULTIPLE_ALIGNMENTz!!AA_MULTIPLE_ALIGNMENTPileUpz is not a known GCG MSF header: z,  zMSF: z..z6Reached end of file without MSF/Type/Check header linezMSF:zType:)Check:z CompCheck:zsGCG MSF header line should be ' MSF: Type: Check: ..', not: %rzCGCG MSF header line should have MSF: for column count, not %r)PNz]GCG MSF header line should have 'Type: P' (protein) or 'Type: N' (nucleotide), not 'Type: %s'z//zName: zName:zLen:zWeight:rzMalformed GCG MSF name line: zDuplicated ID of z4End of file while looking for end of header // line.z.End of file after // line, expected sequences.z4After // line, expected blank line before sequences.zUnexpected line 'z ' in input-z-Received longer sequence than expected for %sc3&K|] }|dk( yw)rN).0lengths E/home/agent/.friday_env/lib/python3.12/site-packages/Bio/Align/msf.py z9AlignmentIterator._read_next_alignment..s766Q;7sz*End of file where expecting sequence data.c32K|]}t|yw)N)len)rseqs rrz9AlignmentIterator._read_next_alignment..s.#SX.sz6GCG MSF headers said alignment length %i, but found %i) stacklevel~.weight)idname description annotations)readlinetell ValueErrorstripsplitjoinrstripendswithindexint startswithfloatappendrisdigitcountallmaxwarningswarnr enumeratereplaceencoderparse_printed_alignmentziprrr)selfstreamline known_headerspartsoffset aln_lengthseq_typenames remainingchecksweightswords index_name index_len index_weight index_checkr%rr#checkseqsr0wordr coordinatesrecords alignmentcolumnss r_read_next_alignmentz&AlignmentIterator._read_next_alignments {{}! /T9 X  ::<    "- 7::<%%'*DIIm,DF   WD;;t$D$4==#6 W UV VN V$ !  '59FG &  UD::  #4TF*!EFDP QR.  s=R6ASS4T-&T26SS14T*T*T*N)__name__ __module__ __qualname____doc__fmtrXrrr r s% CVr^r ) r\r9Bior Bio.AlignrrBio.Seqr Bio.SeqRecordrr rr^rrcs2& #[ 44[r^