o Rŀg3%@s^dZddlZddlmZddlmZddlmZddlmZddl m Z Gdd d ej Z dS) 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) SeqRecordc@seZdZdZdZddZdS)AlignmentIteratorzGCG MSF alignment iterator.MSFc Cszt|}Wnty|dkrtddYdSwgd}|d|vr;td|dd|f|D]}|d}d|vrO|drOnq=td |}| d }||d d ksk|d dvrqtd|z t ||d}Wntytd||ddw||d}|dvrtd|g}g} g} g} |D]x}|}|dkrnq| dr |} z| d} | d}| d}| d}Wntytd|dw| | d}t | |d}t | |d}| |d}||vr td|| || || || |qtdzt|}Wn ty8tddw|rBtdd gt|}|D]r}|} | sVqK| d}z| |}Wn!ty| D] }|stnqkYqKtd!|d"dwd | dd}| |t||d#}|dkrtd$||||7<|| |<td%d&| DrnqKtd'|D] }|d ksJqtd(d&|D}||krtjd)||ftd d*|}t|D]'\}}d |d+d#d,d#}t||kr|d#|t|7}|||<qd-d.|D}t|\}}d/d.t||| D}t||}|j}||krGtd)||f|S)0Nrz Empty file.)z!!NA_MULTIPLE_ALIGNMENTz!!AA_MULTIPLE_ALIGNMENTPileUpz$%s is not a known GCG MSF header: %sz,  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:r zMalformed 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 %scss|]}|dkVqdS)rN).0lengthrrA/var/www/html/myenv/lib/python3.10/site-packages/Bio/Align/msf.py z9AlignmentIterator._read_next_alignment..z*End of file where expecting sequence data.css|]}t|VqdS)N)lenrseqrrrrrz6GCG MSF headers said alignment length %i, but found %i) stacklevel~.cSsg|]}|qSr)encoderrrr sz:AlignmentIterator._read_next_alignment..c Ss,g|]\}}}tt||||d|idqS)weight)idname description annotations)rr)rr$rr"rrrr!s)next StopIterationtell ValueErrorstripsplitjoinrstripendswithindexint startswithfloatappendrisdigitcountallmaxwarningswarnr enumeratereplacerparse_printed_alignmentzipr)selfstreamline known_headerspartsoffset aln_lengthseq_typenames remainingchecksweightswords index_name index_len index_weight index_checkr$rr"checkseqsr0wordr coordinatesrecords alignmentcolumnsrrr_read_next_alignments     '                            z&AlignmentIterator._read_next_alignmentN)__name__ __module__ __qualname____doc__fmtrWrrrrrs r) r[r9Bior Bio.AlignrrBio.Seqr Bio.SeqRecordrrrrrrs