ó ZIc@sdZddlmZddlmZddlmZddlmZddl m Z dd„Z defd „ƒYZ ed krdd lmZd GHHd GHedƒZxFe eƒD]8ZeGHx*eD]"ZeejƒGejGejGHqÊWq¸WdGHHdGHedƒZxFe eƒD]8ZeGHx*eD]"ZeejƒGejGejGHq*WqWdGHHdGHdeee eƒƒƒƒks†t‚dGHHdGHeƒZe eƒjegƒejdƒejƒGHeƒZy/e eƒjeegƒesÿtdƒ‚Wqek rqXndS(sj Bio.AlignIO support for the "nexus" file format. 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). See also the Bio.Nexus module (which this code calls internally), as this offers more than just accessing the alignment or its sequences as SeqRecord objects. iÿÿÿÿ(tNexus(t Alignment(t SeqRecord(tAlignmentWriter(tAlphabetc cs tj|ƒ}|js!t‚nt|jƒ}t|jƒt|jƒksTt‚|r‘|t|jƒkr‘t dt|jƒ|fƒ‚nxmt |j|jƒD]V\}}|j |ƒsÈt‚|j|}|j j t|d|d|ddƒƒq§W|VdS(sFReturns SeqRecord objects from a Nexus file. Thus uses the Bio.Nexus module to do the hard work. You are expected to call this function via Bio.SeqIO or Bio.AlignIO (and not use it directly). NOTE - We only expect ONE alignment matrix per Nexus file, meaning this iterator will only yield one Alignment.s$Found %i sequences, but seq_count=%itidtnamet descriptiontN(Rtmatrixt StopIterationRtalphabettlentunaltered_taxlabelst taxlabelstAssertionErrort ValueErrortzipt startswitht_recordstappendR(thandlet seq_counttnt alignmenttold_nametnew_nametseq((s†/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/AlignIO/NexusIO.pyt NexusIterators   $" t NexusWritercBs)eZdZd„Zd„Zd„ZRS(sÔNexus alignment writer. Note that Nexus files are only expected to hold ONE alignment matrix. You are expected to call this class via the Bio.AlignIO.write() or Bio.SeqIO.write() functions. cCsœt|ƒ}y|jƒ}Wntk r5d}nX|dkrFdSy|jƒ}Wntk rod}nX|dk r‹tdƒ‚n|j|ƒdS(sÒUse this to write an entire file containing the given alignments. alignments - A list or iterator returning Alignment objects. This should hold ONE and only one Alignment. Ns0We can only write one Alignment to a Nexus file.i(titertnextR tNoneRtwrite_alignment(tselft alignmentst align_itertfirst_alignmenttsecond_alignment((s†/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/AlignIO/NexusIO.pyt write_fileHs        cCs¾t|jƒƒdkr'tdƒ‚n|jƒdkrHtdƒ‚ndd|j|jƒ}tj|ƒ}|j|_x*|D]"}|j|j |j j ƒƒq„W|j |j ƒdS(NisMust have at least one sequences Non-empty sequences are requireds.#NEXUS begin data; dimensions ntax=0 nchar=0; sformat datatype=%s; end;(R t get_all_seqsRtget_alignment_lengtht_classify_alphabet_for_nexust _alphabetRR t add_sequenceRRttostringtwrite_nexus_dataR(R"Rtminimal_recordRtrecord((s†/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/AlignIO/NexusIO.pyR!cs   cCs‚tj|ƒ}t|tjƒs0tdƒ‚nNt|tjƒrFdSt|tjƒr\dSt|tjƒrrdStdƒ‚dS(svReturns 'protein', 'dna', 'rna' based on the alphabet (PRIVATE). Raises an exception if this is not possible.sInvalid alphabettproteintdnatrnas#Need a DNA, RNA or Protein alphabetN(Rt_get_base_alphabett isinstancet TypeErrortProteinAlphabett DNAAlphabett RNAAlphabetR(R"R ta((s†/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/AlignIO/NexusIO.pyR*ss(t__name__t __module__t__doc__R'R!R*(((s†/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/AlignIO/NexusIO.pyR?s  t__main__(tStringIOsQuick self tests#Repeated names without a TAXA blocksù#NEXUS [TITLE: NoName] begin data; dimensions ntax=4 nchar=50; format interleave datatype=protein gap=- symbols="FSTNKEYVQMCLAWPHDRIG"; matrix CYS1_DICDI -----MKVIL LFVLAVFTVF VSS------- --------RG IPPEEQ---- ALEU_HORVU MAHARVLLLA LAVLATAAVA VASSSSFADS NPIRPVTDRA ASTLESAVLG CATH_HUMAN ------MWAT LPLLCAGAWL LGV------- -PVCGAAELS VNSLEK---- CYS1_DICDI -----MKVIL LFVLAVFTVF VSS------- --------RG IPPEEQ---X ; end; tDones Repeated names with a TAXA blocksO#NEXUS [TITLE: NoName] begin taxa CYS1_DICDI ALEU_HORVU CATH_HUMAN CYS1_DICDI; end; begin data; dimensions ntax=4 nchar=50; format interleave datatype=protein gap=- symbols="FSTNKEYVQMCLAWPHDRIG"; matrix CYS1_DICDI -----MKVIL LFVLAVFTVF VSS------- --------RG IPPEEQ---- ALEU_HORVU MAHARVLLLA LAVLATAAVA VASSSSFADS NPIRPVTDRA ASTLESAVLG CATH_HUMAN ------MWAT LPLLCAGAWL LGV------- -PVCGAAELS VNSLEK---- CYS1_DICDI -----MKVIL LFVLAVFTVF VSS------- --------RG IPPEEQ---X ; end; sReading an empty fileis Writing...s-Should have rejected more than one alignment!N(R=t Bio.NexusRtBio.Align.GenericRt Bio.SeqRecordRt InterfacesRtBioRR RRR;R?RR:trtreprRRRR tlistRR'tseektreadtFalseR(((s†/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/AlignIO/NexusIO.pytsX $H   $  $'