{Ic @sXdZddlZddlZddlZddlZddlZddlmZddlm Z ddl m Z ddl m Z mZyddlZWnek reZnXeZy eZWn!ek rddlmZnXdZd d d d d ddddddddg ZddddddgZdZdZdZdgZd ZdZ d!Z!d"Z"d#e#fd$YZ$d%d=d&YZ%d'd>d(YZ&ed)Z'd*Z(d+d,gd-Z)d.Z*d/Z+d0Z,d1Z-d2Z.d3Z/d4Z0d5Z1d6Z2d7d?d8YZ3d9d@d:YZ4d;e5fd<YZ6dS(AsNexus class. Parse the contents of a nexus file. Based upon 'NEXUS: An extensible file format for systematic information' Maddison, Swofford, Maddison. 1997. Syst. Biol. 46(4):590-621 iN(tIUPAC(t IUPACData(tSeq(tTreetNodeData(tSetiFtcharstatelabelst charlabelst taxlabelsttaxsettcharsett charpartitiont taxpartitiontmatrixttreetutreet translatet codonpossetttitlettreestdatat charactersttaxatsetstcodonss()[]{}/\,;:=*'"`+-<>t?abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890_s t&tcharstcodonpositionssH#NEXUS begin data; dimensions ntax=0 nchar=0; format datatype=dna; end; t NexusErrorcBseZRS((t__name__t __module__(((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyR3st CharBuffercBsheZdZdZdZdZdZdZdZdZ dZ d Z d Z RS( s7Helps reading NEXUS-words and characters from a buffer.cCs%|rt||_n g|_dS(N(tlisttbuffer(tselftstring((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt__init__7scCs|jr|jdSdSdS(Ni(R"tNone(R#((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytpeek=s  cCs.dj|jj}|r&|dSdSdS(Nti(tjoinR"tstripR&(R#tb((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytpeek_nonwhitespaceCscCs!|jr|jjdSdSdS(Ni(R"tpopR&(R#((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytnextJs cCs=x6tr8|j}|dkr%Pn|tkr|SqWdS(N(tTrueR.R&t WHITESPACE(R#tp((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytnext_nonwhitespacePs    cCs.x'|jdtkr)|jd|_qWdS(Nii(R"R0(R#((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytskip_whitespaceYscCsmxf|D]Z}y|jj|}Wntk r6qXdj|j| }|j||_|SqWdSdS(NR((R"tindext ValueErrorR)R&(R#ttargettttpostfound((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt next_until]s  cCs dj|jt| |kS(NR((R)R"tlen(R#tword((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt peek_wordjscCsg}t}|j}|s"dS|j||dkrDt}n|tkrT|Sxtr |j}|dkr|j|j|jdkr|j}q|rPqqW|r|j|jqW| s|tks|tkrPqW|j|jqWWdj |S(sReturn the next NEXUS word from a string. This deals with single and double quotes, whitespace and punctuation. t'R(N( tFalseR2R&tappendR/t PUNCTUATIONR'R.R0R)(R#R<tquotedtfirsttctskip((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt next_wordms0        cCsdj|jS(s.Return the rest of the string without parsing.R((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/Nexus/Nexus.pytrests( RRt__doc__R%R'R,R.R2R3R:R=RFRG(((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyR 5s       t StepMatrixcBsPeZdZdZdZdZdZdZdZddZ RS( sbCalculate a stepmatrix for weighted parsimony. See Wheeler (1990), Cladistics 6:269-275. cCsi|_g|D] }|^q|_|jj|rK|jj|nxW|jD]L}xCg|jD]}||krh|^qhD]}|j||dqWqUWdS(Ni(RtsymbolstsortR@tset(R#RJtgaptstxty((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyR%s  /cCs1||kr||}}n||j||s(treduceRtvalues(R#((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytsumscCsQ|j}|dkrMx2|jD]$}|j|t||j|s''(treplaceR)t MRBAYESSAFERLt intersectionR0RA(R`tmrbayestsafeRD((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytsafenames 1cCs^|s dSxM|jdr+|jdsI|jdrY|jdrY|dd!}q W|S(s6Remove quotes and/or double quotes around identifiers.R>t"iiN(R&t startswithtendswith(R<((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt quotestrips ?t-t?cCst|}|dkrdS|d}x*|dkrR|||krR|d8}q)Wd}x*||kr|||kr|d7}q\W||kr|dkrdS||fSdS(skReturn position of first and last character which is not in skiplist. Skiplist defaults to ['-','?']).iiiN(NN(ii(R;R&(tsequencetskiplisttlengthtendtstart((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt get_start_ends   cCsEg|D] }||r|||f^q}|jt|dS(s9Returns a sorted list of keys of p sorted by values of p.i(RKtzip(R1tpntstartpos((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt_sort_keys_by_valuess- cCs tt|}|j|S(sMCheck that all values in list are unique and return a pruned and sorted list.(R!RLRK(tl((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt _make_uniques cCsx||kr|jddjdrwdj|jdd dttd|jdddd}q|d7}qW|S(s=Returns a unique name if label is already in previous_labels.t.itcopys.copyt0ii(tsplitRmR)tstrteval(tprevious_labelstlabel((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt _unique_label s LcCs-tg|D]}|||jf^q S(s?Converts a Seq-object matrix to a plain sequence-string matrix.(tdictttostring(R R7((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt_seqmatrix2strmatrixscCs|s dStt|}|jg}|}|j|ddxKt|dkrd}x,t|D]\}}||d||krqpqp|dkrt|dkr||d|||dkr||d}qp|| }t|dkr#|jt|ddn`|dkrY|jd|dd|ddfn*|jd|dd|dd|f||}PqpWqKWd j|S( sTransform [1 2 3 5 6 7 8 12 15 18 20] (baseindex 0, used in the Nexus class) into '2-4 6-9 13-19\3 21' (baseindex 1, used in programs like Paup or MrBayes.). R(ig?iiis%d-%ds%d-%d\%dRd(R!RLRKR@R;t enumerateRR)(t orig_listt shortlisttclisttsteptiROtsub((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt_compact4nexuss. >  **  c Cs|s dS|dd}tj|dd}ttg|D]}|dj^q<dk}|rsd|_nd|_d|_t|_ d|_ x>|j j D]-\}}||j d||f<|j |=qWx>|j j D]-\}}||j d||f<|j |=qWiit|j|6d6|_x!|dD]\}} g|jD]} | | jkrZ| ^qZ} g|jD]} | | kr| ^q} g| jD]} | | kr| ^q} x^| D]V} |j| ct| j| jj| j|jj| j|j|j7t[it]s#Nexus formatting error: unmatched ]t;s s#Nexus formatting error: unmatched [N( R R?R/R.R&R'tSPECIALCOMMENTSRR@R)(ttexttcontentstnewtexttnewlinet quotelevelt speciallevelt commlevelR7((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt_kill_comments_and_break_lines{sF     , %        cCsg}x|D]w}|jddjddj}|jjdr\|j|q |jdd}|r |j|q q W|S(sAdjust linebreaks to match ';', strip leading/trailing whitespace. list_of_commandlines=_adjust_lines(input_text) Lines are adjusted so that no linebreaks occur within a commandline (except matrix command line) s s s R Rd(RfR*tlowerRmR@(tlinestformatted_linesR|((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt _adjust_liness $cCs |jd}x|dkr|jd}|dkrLtd|n||krktd|ng||d|!D] }|^q}}|jdj|}||j}||jkr|j}n|| |||d}|jd}qW|S( s>Replaces ambigs in xxx(ACG)xxx format by IUPAC ambiguity code.t(it)is Missing closing parenthesis in: s Missing opening parenthesis in: iR((tfindRRKR)tupperR(tseqtrev_ambig_valuestopeningtclosingROtambigt ambig_code((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt_replace_parenthesized_ambigss   $ t CommandlinecBseZdZdZRS(s/Represent a commandline as command and options.c Cs$i|_g}d|_y%|jjdd\|_}Wn=tk r||jd|_dj|jd}nX|jjj|_|jtkr|j|_njt |dkr y1|j ddj}gt t |D]K}||dkr|dkr|t |kr|d||df^q}g}x|D]}|j |qTWgt t |D]}||kr~|^q~}x1|D])} || d|j|| djiiRdR(RRt nucleotidetrnatproteintDNtBtEQtZtXt*tstandards'Datatype standard is not yet supported.sUnsupported datatype: it1RRRMRRRRsTRANSPOSE is not supported!RtyesRtnotokens(RRR(.R/RRJRmRntsymboldRfRRR!RLRRRRt ambiguous_dnaRRtambiguous_dna_valuestambiguous_valuestunambiguous_dna_letterstunambiguous_letterst ambiguous_rnatambiguous_rna_valuestunambiguous_rna_lettersRtprotein_lettersRR)tkeystvalid_charactersRRtrev_ambiguous_valuesRKtlettersRRMRRRRR&RRR?(R#RRtrevRYtvRDtkey((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt_formats    $$"  *" "F            &   cCs ||_dS(N(RL(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/Nexus/Nexus.pyt_setscCs ||_dS(N(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/Nexus/Nexus.pyt_optionsscCs ||_dS(N(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/Nexus/Nexus.pyt _eliminatescCsdS(s&Get taxon labels (PRIVATE). As the taxon names are already in the matrix, this is superfluous except for transpose matrices, which are currently unsupported anyway. Thus, we ignore the taxlabels command to make handling of duplicate taxon names easier. N((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/Nexus/Nexus.pyt _taxlabelsscCsStg|jD]}|jdd|f^q }|jdd}|j|S(s.Check for presence of taxon in self.taxlabels.RdRe(RRRftget(R#ttaxonR7tnextaxatnexid((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt_check_taxlabelss4cCsi|_t|}xtry|j}|dkr=Pn|j|dt}t|j}||j|<|j}|dkrPn|dkrt d|nWqt k rqt d|qXqWdS(Ntset_typet,sMissing ',' in line %s.sFormat error in line %s.( RR R/RFR&t_resolvetCHARSETRoR2R(R#Rtoptstwt identifiertstateRD((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt _charlabelss&          cCs|j|dS(N(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/Nexus/Nexus.pyt_charstatelabels.scCsdS(N((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/Nexus/Nexus.pyt _statelabels2sc Cs^|j s|j r#tdni|_d}t}g|jdD]$}|jdkrH|j^qH}t|}x8y|j}WnNt k r||jkrtdq||jkrtdqPnX|d7}||jkr"|j stdq"d}t }nt |}t |j}|jj}d} |j r|rdj|j} qdj|jj} nVdj|j} x>t| |jkr|j}| dj|j7} qWtt| |j|j} |dkr&| } nq|jrxe| jj|j} | d krZPnt| j| | | | j| d|j} q2Wnxxt| jD]d\} }||jkr||jkr||jkrtd |||| d | d !fqqW|r`|jj|t|jj |}| |j|<|j!j|qt|j!|d |}|j"|}|r|j|c| 7            - ;$cCsStd}||_x'|D]}|jjt||qW|jj|dS(NR((RRRR@RR(R#RRR R((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyR s    cCs2|j|dt\}}t||j|( R?R t_name_n_vectorRR/R.R&R\R]R}R( R#RR RR:R`RR;tsubnamet subindices((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt _taxpartitions&   !   cCs|jj}|j|g|jjD]}||kr,|^q,}|gksht|dkr{td|n |d|_dS(sRead codon positions from a codons block as written from McClade. Here codonposset is just a fancy name for a character partition with the name CodonPositions and the partitions N,1,2,3 is$Formatting Error in codonposset: %s iN(RR%t_charpartitionR;RR(R#Rtprev_partitionsRt codonname((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt _codonpossets  .cCsdS(N((R#R((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt_codesetsc Csi}t}t|}|j|}|s@td|nd}xtr|j}|dksz|dkr| r|j|dtdd\}} t | ||( R?R RcRR/R.R&R\R9R}R( R#RR RR:R`RR;RdRe((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyRgs&   !   RcCs_t|}|j|d|}|j|d|}|dkrUtd|n||fS(sParse the taxset/charset specification (PRIVATE). e.g. '1 2 3 - 5 dog cat 10 - 20 \ 3' --> [0,1,2,3,4,'dog','cat',9,12,15,18] RaR6sFormatting error in line: %s N(R Rct _parse_listR&R(R#RR6RaR:R`R((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyR\s   cCs|j}|j}|dkr3|j}n|sLtd|nt|}|jdkr|j}|j}|j}|jdkrtdtq|jdkrtd|tfqn|j|krtd|n|S(s6Extract name and check that it's not in vector format.RsFormatting error in line: %s Rtvectors$Unsupported VECTOR format in line %sRsUnknown qualifier %s in line %s(RGRFRRoR,R2RR(R#R:RaRGR`Rt qualifiertclose((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyRc*s(        c Csg}|jryxtr|j}|s4Pn|j|d|}|jdkrl|}d}|j}|j|jd|}|tkr|jdkr|j} t|j}n|jt||d|qt |t kst |t kr$t d|n|j j |}|j j |}|j ||d!} |j| qt |t kr|j|q|j|qWWqt k rqdSXn|S(sParse a NEXUS list (PRIVATE). e.g. [1, 2, 4-8\2, dog, cat] --> [1,2,4,6,8,17,21], (assuming dog is taxon no. 17 and cat is taxon no. 21). R6Rpis\s:Name if character sets not allowed in range definition: %sN(R,R/RFR8R2R9RNRRttypeR!RRR4R@R&( R#toptions_bufferR6t plain_listR<RvRuRthyphent backslashttaxrange((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyRlBsB       $  cCst|}|s!tdn|tkryt|}Wntk r|jr||jjkrx^|jD]}|j||kru|SquWq|jr||jkr|j|Std|qX||jkr|dStd||jfn|t kryt|}Wn]tk r~|j |}|rH|S|j rk||j krk|j |Std|qX|dkr||j kr|j |dStd||j fntd|d S( s!Translate identifier in list into character/taxon index. Characters (which are referred to by their index in Nexus.py): Plain numbers are returned minus 1 (Nexus indices to python indices) Text identifiers are translaterd into their indices (if plain character indentifiers), the first hit in charlabels is returned (charlabels don't need to be unique) or the range of indices is returned (if names of character sets). Taxa (which are referred to by their unique name in Nexus.py): Plain numbers are translated in their taxon name, underscores and spaces are considered equal. Names are returned unchanged (if plain taxon identifiers), or the names in the corresponding taxon set is returned s0INTERNAL ERROR: Need type to resolve identifier.s Unknown character identifier: %sis-Illegal character identifier: %d>nchar (=%d).sUnknown taxon identifier: %sis(Illegal taxon identifier: %d>ntax (=%d).sUnknown set specification: %s.N(RoRR9RNR5RRURRR]R5RRR(R#R<R6RRYt taxlabels_id((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyR8nsF        cCsdS(N((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/Nexus/Nexus.pyt _statesetscCsdS(N((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/Nexus/Nexus.pyt _changesetscCsdS(N((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/Nexus/Nexus.pyt_treesetscCsdS(N((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/Nexus/Nexus.pyt_treepartitionsR(c Cs|s|j}n|sdS|s.|j}n|r<i} x|D]} g|} | jgt|jD]} | || krg| ^qgt| } |d| d}|jd}|dkr|| d| d}n|d| }|| | <|jd|d |d |d |d | d |d|dtd| qAW| S|jd}|jd|d |d |d |d |d |d|dtd| |SdS(sWrites a nexus file for each partition in charpartition. Only non-excluded characters and non-deleted taxa are included, just the data block is written. Ns Partition: s R~iRes.dataRR t blocksizeRtexcludetdeletetcommentt append_setsRi( R RRRRR}trfindtwrite_nexus_dataR?(R#R RR{RR|R}R R~Rit pfilenamesR1t total_excludeRDtpcommenttdott pfilenametfn((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytwrite_nexus_data_partitionss6    9     !c  Cs|s|j}n|sdS|s.|j}ng|D]} |j| s5| ^q5rtddjt|jt|jn|r||jkrtd|q|j|}t |}i}x>|D]3}g||D]}||kr|^q||}||krrdt|d j t t||f^qrfqqW|jdt|dkrdSdj |SdS(sReturns a sets block.R(s begin codonss begin setsiiischarset %s = %sRistaxset %s = %sRdRR s %s %s = %ss, s%s: %sstaxpartition %s = %ssend; is; N(RRRRRR@RRkRR)tCODONPOSITIONSR{RtmapR;(R#R|R}RiRRtsetsbtoffsettofflistRDRtnstcsetRNR7ttsetR1RRtsntnspR((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyRbsh    )*10  - N  )` iFcCs|smd|jkr]|jjddjd kr]dj|jjdd d}qm|jd}nt|d}x|jD]z}|jd t|d xVtd t |j |j |D]/}|j|j |j |||!d qWqW|j |S( sAWrites matrix into a fasta file: (self, filename=None, width=70).R~itpauptnexustnextdats.fasR;t>s i(RRRR( RRRR)RRRRkRR;R RRo(R#RtwidthRR2R((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt export_fastas1&,1 cCs|smd|jkr]|jjddjd kr]dj|jjdd d}qm|jd}nt|d}|jd |j|jfx;|jD]0}|jd t ||j |j fqW|j |S( sWrites matrix into a PHYLIP file: (self, filename=None). Note that this writes a relaxed PHYLIP format file, where the names are not truncated, nor checked for invalid characters.R~iRRRRs.phyR;s%d %d s%s %s (RRRR( RRRR)RRRRRRkR RRo(R#RRR2((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt export_phylips1&. cCs|s|j}ng|jD]$}||kr||kr|^q}|sPdSt|dkrgtt||dD]}||kr}|^q}Sgt||djD]?\}}||kr||jj|j |j f^q}x}|dD]q} g} x\|D]T} || | dj } | |j ks|| |j krl|j dj dks|| | dkr| j| q| | dks| d|j ks|j dj dkr| d|j kr| j| d|jj| | fq| |jkrtj|j| jtj| d} | rk| j| ddj| fqkqqW| }qWg|D]}|d^q}|S(s+Return a list with all constant characters.iiRRR(N(R RR&R;RRRRR1RRRMRRR@RRLRhR)(R#R R}R|R7RRORtconstantR2t newconstanttsitetseqsitet intersectRNtcpos((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyRs0 47] GO),* c CsZg|jD]}||kr |^q }|s2dSg}x|D]}|j||j}|jjddkr||jkr|j}n|r||jkr||kr|j|qq?|j g|j |D]}||kr|^qq?W|j|krL|rLt |dkrLg|D]}||jkr(|^q(}n|j |S(sSummarize character. narrow=True: paup-mode (a c ? --> ac; ? ? ? --> ?) narrow=false: (a c ? --> a c g t -; ? ? ? --> a c g t -) RRiN( RR&R RRR1RMRR@RRR;RK( R#RR}tnarrowR7RtcstatusRDR+((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyRs ( '  7'+ R^c Cst|j|j}xgt|jD]}||kr(|^q(D]q}|j||}xVt|d D]D\}} x5||dD]%} |j| j| jdqWqmWqDW|j j j d|S(sCalculates a stepmatrix for weighted parsimony. See Wheeler (1990), Cladistics 6:269-275 and Felsenstein (1981), Biol. J. Linn. Soc. 16:183-196 iiR`( RIR RMRRRRRRRRZR]Rc( R#R`R|R}RRNRRRtb1tb2((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytweighted_stepmatrixs5+c Cs|s|j}ng|D]}|j|s|^qrktddjtj|j|jn|gkrg|jD]$}||kr||kr|^q}|siSg|D]}||j^q}t |}gt |D]\} } | |kr| ^q} | gkrPt g|D]}|t d|j f^q+St | } gtdj| D]} t | |j ^qo}t t ||SnBt g|jD].}||kr||kr|||f^qSdS(s=Return a matrix without deleted taxa and excluded characters.sUnknown taxa: %ss, R(N(R R5RR)RRLRRRRxRRRRR( R#R R}R|R7RRYRtzipped_mRRNtsitesmt zipped_sitesm((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyRs$ (+ 4# 1 / 1cCs|s|j}nt||jdt}|jd|d|}|sQiSt||jddkru|Sg|jD]}||kr|^q}|rtg|D]}||j^q}||jdj } n#tg|D]}||^q}gt t|D]&} |t j dt|d^q} t djt| } |rg| D]} t| | ^qm} ntt|| S(sReturn a bootstrapped matrix.iR}R|iR((R RR%RRR;RRxRRRtrandomtrandintRR)R(R#R R}R|t seqobjectstcmR7RRRRtbootstrapsitesmt bootstrapseqsRN((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt bootstrap*s$  ()#?%cCs|stdn|jt|}|dkrK|j|j| n |dkrk||j|7}n||jkrt|j|}n|}||jjkst dt ||j |j|<|j d7_ |jj ||jj |dS(s'Adds a sequence (string) to the matrix.sNew sequence must have a nameisVERROR. There is a discrepancy between taxlabels and matrix keys. Report this as a bug.iN(RRR;t insert_gapRRRR R%RBRRRR@R(R#R`Rrtdifft unique_name((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyt add_sequence@s  !icCstd}|dks'||jkr:td|n|dkrJdS|jrtg|jD]}|j|j^q`}ng}dgt|jg||||+t|}tdj |}gt |jD](\} } | t || |j f^q} t | |_|j|7_x<|jjD]+\} } || ||d||j| # +cCsA|r|rtdn|js(dS|jj}|ji}|r|j|jtjd}x|D]K}||krtx|||kr|d7}qW|j||||RRt __slots__R&R%RRtpropertyRRRRRR R R,R-R.R/R0R5R>R?R@RMRORQRPR R^R`RfRjRkRgR9R\RcRlR8RwRxRyRzR?RR/RRRRRRRRRRRRRRR(((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pyRsv  (   7      Q         Y   "         , 8     %    E & 5  (((((7RHRRR[RRt Bio.AlphabetRtBio.DataRtBio.SeqRtTreesRRRt ImportErrorR?RR/RLt NameErrorRRt INTERLEAVERRRARgR0RR9R]RRt ExceptionRR RIRkRoRwR{R}RRRRRRRRRtobjectR(((s/oak/stanford/groups/akundaje/marinovg/programs/biopython-1.50.tar.gz/biopython-1.50/build/lib.linux-x86_64-2.7/Bio/Nexus/Nexus.pytsX<     \?      " C <  "