;Wc@sddlmZddlZddlZddlZddlZddlZddlZddlZddl m Z ej ddddgZ dZ d Zd Zd Zd Zd ZdZdefdYZdefdYZdefdYZdefdYZdefdYZdZdZdZeeefZedZedZ dZ!dZ"e#d kre"ndS(!i(twith_statementN(t quoteattrt LabColourtltatbcCs!t|t|kr xtj||D]\}}|jr||jru||kr_dS||kr+dSq+qdSn|jrdS|jr|jr||krdS||kr+dSq+qdSn|jrdS||krdS||kr+dSq+WdSt|t|SdS(s&Compares 2 strings of alphabet symbolsiiiN(tlent itertoolstiziptisalphatisdigit(txtyRR((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytalphabetSymbolCompare s8            c Cs8|d?d@d}|d?d@d}|d@d}t|||}t|||}||}||d}d}d} ||kr+|dkr|d||} n|||} ||kr|||}||kr|d7}qn3||kr |||d}n|||d}|d:}n|| |fS( sCreate HSL from an RGB integeriiiiig?ii(tmintmax( trgbtrtgRtmin_ctmax_ctdelta_cRthts((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt_rgb2hsl/s,       c Csd}|d}|d}|d}|dkrH|}|}|}n|dkre|d|}n||||}d||} || ||dd}|| ||}|| ||dd}tt|dd >tt|dd >Btt|dB} | S( s&Create an RGB integer from a HSL tuplecSs|dkr|d7}n|dkr2|d8}n|ddkrV|||d|S|dkrf|S|ddkr|||dd|dS|SdS(Niig?g@g?g@g@((tptqtt((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt_hueLs     iiig?g?g@iii(tinttround( thslRRRRRRRRRR((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt_hsl2rgbJs"      FcCsEt|}|d|d|dd|dddf}t|S(s$Make a lighter version of the colouriiig?i(RR (RR((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt_lightenms /cCs|}|}|}|dkr1|d}tj|}||}|d|} |d||} |d|d|} |dkr|}| }| }q1|dkr| }|}| }q1|dkr| }|}| }q1|dkr| }| }|}q1|dkr| }| }|}q1|}| }| }ntt|d} tt|d} tt|d}| d >| d >B|B}|S( sCreate an RGB integer from HSVii<g?iiiiiii(tmathtfloorRR(thuetsattvalueRRRRtitfRRRtredtgreentblueR((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt_hsv2rgbssL              c Csd}d}||d?d@}||d?d@}||d@}|d|d|d}|d |d |d }|d |d |d}||d}||d}||d}d|d} d||} d||} t| | | S(s#Convert an RGB integer to Lab tuplecSsQ|d}|dkr9|dd}tj|d}n |d:}|d9}|S(s.Helper function for XYZ colourspace conversionigbX9Ȧ?g)\(?gzG?g333333@gףp= )@id(R"tpow(R&tc((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt xyzHelpers    cSsA|}|dkr+tj|dd}nd|dd}|S(s.Helper function for Lab colourspace conversiong2#?g?g@gS%@g0@g]@(R"R-(R&R.((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt labHelpers  iiig7d?g,C?gv?gz6?g,C?g]m{?gN@aÓ?g2%䃾?g"~j?g^I W@gY@gn8[@g]@g@@gi@(R( RR/R0tc1tc2tc3R R tzRRR((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt_rgb2labs c Cstj|j|j|j|j}tj|j|j|j|j}||}|j|j}|j|j}|j|j}d}||||||} | dkrtj| }n|} |dd|} |dd|} tj| | | | | | S(s;Calculate the distance between 2 colours in Lab colourspaceig?g ףp= ?gQ?(R"tsqrtRRR( tlab1tlab2R1R2tdctdltdatdbtdht dh_squaredtfirsttsecondtthird((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt _lab_dists''  tAlphabetParseErrorcBs eZdZdZdZRS(s%An error in the alphabet being parsedcCs ||_dS(N(R&(tselfR&((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt__init__scCs t|jS(N(treprR&(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt__str__s(t__name__t __module__t__doc__RERG(((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyRCs tAlphabetReaderSymbolcBs8eZdZdddZdZdZdZRS(s4Temporary information storage for an alphabet readercCsC||_||_||_||_||_g|_t|_dS(N(tsymboltnametcolourt complementtcomprisetaliasestFalsetalias(RDRLRMRNRORP((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyREs      cCs|dkrdSt|ts#dS|jdksGt|jdkr|jdkskt|jdkr~t|j|jSdSn;|jdkst|jdkrdSt|j|jSdS(Nii(tNonet isinstanceRKRPRR RL(RDtobj((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt__cmp__s $$$cCs t|jS(N(RFRL(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyRGscCs"|jdko!t|jdkS(Ni(RPRTR(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt isAmbiguoussN(RHRIRJRTRERWRGRX(((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyRKs   tAlphabetReadercBseZdZdZdZdZdZdZdZdZ d d dZ d d d d d Z d Z d Zd ZRS(sEReads an alphabet and validates it before creating an alphabet objectcCsd}d}d}d|d|d|d}tjd|d |_tjd |d |_tjd |d |d |_tjd |d |d|_t|_t|_t|_ t|_ t|_ t|_ t|_ d|_d|_i|_g|_g|_dS(Ns[A-Za-z0-9?.*-]s[0-9a-fA-F]{6}s0("(?:[^\\"]+|\\["\\/bfnrt]|\\u[0-9A-Fa-f]{4})*")t(s)(?:\s+s )?(?:\s+(s))?s^\s*ALPHABET(?:\s+v1)?(?:\s+s$)?(?:\s+(DNA|RNA|PROTEIN)-LIKE)?\s*$s^\s*s\s*$s\s*~\s*s\s*=\s*(s*)\s*$(tretcompilet header_retcore_single_ret core_pair_retambig_reRRtparsedt seen_headert seen_symbolt seen_ambigtseen_lctseen_uctfully_complementableRTRMtliket sym_lookupt core_symst ambig_syms(RDtqr_symt qr_colourtqr_nametqr_core((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyREs(""           cCst}x|jD]}|jdkr1t}qn|j|j}|dkrrtd|jd|jdn|j|jkrtdqqW||_ dS(s8Ensure all referenced complement symbols actually exist.s core symbol s has complement s which has not been defineds+Alphabet symbol complement is not symmetricN( tTrueRjRORTRRRiRCRLt RuntimeErrorRg(RDRgtsym_objtcomplement_sym_obj((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt_checkComplementss %cCsi}xG|jD]<}|j|kr<||jj|q|g||j|D]6}|dkst |j |j dkr|}qqWx<|D]1}||k rt|_|jj|j qqWq]q]Wg|jD]}|js^|^q^|_|j j |jj dS(siCheck every ambiguity character to see if it has the same comprising characters as something elseiiN(RkRPtappendt iteritemsRRiRpRSRQRLRTR Rjtsort(RDtcomprise_lookuptsymRPtsymstcore_symt prime_ambig((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt _mergeAliases's.    '    !( c Cst|jdks:t|jdjt|jkrtdddddjg|jD]}|j^qY}|jjd|||j dt|j|jStdt|dn7||j kr~t|j |jStd|ddS(s$Get the muted colour for this symbols Symbol index s does not exist in alphabetsSymbol N( RURRRRR!RNRqRR(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetMutedColourscCs1t|j|jkr)|j|jjSdSdS(sGet the wildcard symbolN(RRRRLRT(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt getWildcardscCs|jS(sFRetrieve a tuple with all core symbols, immutable membership and order(R(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt getSymbolsscCs|jS(sDRetrieve a tuple with all core symbol complement indicies, immutable(R(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetComplementsscCs ||jkS(s+Check if the symbol is a member of alphabet(R(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt isValidSymbol scCs*||jkr&|j|j|jkStS(s0Check if the symbol is a core member of alphabet(RRRRR(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt isCoreSymbolscCs*||jkr&|j|j|jkStS(s3Check if the symbol is the wildcard of the alphabet(RRRRR(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytisWildcardSymbolscCs<||jkr8|j|jj|jkr8tSntS(sYCheck if the symbol is a member of the alphabet and the normal way to identify the letter(RRLtlowerRpRR(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt isPrimeSymbols"cCs!||jkr|j|jSdS(s,Retrieve the index of the symbol (immutable)N(RRRT(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetIndex"scCs\t|ttfr;|t|jkrX|j|jSn||jkrX|j|jSdS(s?Retrieve the indexes of the comprising core symbols (immutable)N(RURRRRRPRRT(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetComprisingIndexes(s cCs\t|ttfr;|t|jkrX|j|jSn||jkrX|j|jSdS(s?Retrieve the indexes of the comprising core symbols (immutable)N(RURRRRRRRT(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetPairIndexes2s cCs'|t|jkr#|j|jSdS(s$Return the symbol at the given indexN(RRRLRT(RDR((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt getSymbol=scCs*|t|jkr&|j|jSdS(N(RRRRT(RDR((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetXmlIdCscCs{|j|}g}xXtj|j|jD]>}|j||jr,|jr,|j|jq,q,Wdj |S(s/Returns all symbols variants at the given indexR( RRRRLRQRuRR RR(RDRRRzRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt getAliasesHs  cCsnt}x5|D]-}||jkr)dS||j|jO}qWt|}||jkrj|j|jSdS(sFind a symbol that matches the string of comprising symbols. Duplicates and aliases are allowed in the string but all included symbols must be known.N(RRRTRPRRRL(RDRPtsymi_setRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt findSymbolSs   cCs|jS(s$Do all core symbols have complements(Rg(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytisComplementableascCs\t|ttfr;|t|jkrX|j|jSn||jkrX|j|jSdS(s2Retrieve the index of the complement of the symbolN(RURRRRRORRT(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetComplementIndexes cCs-|j|}|dkr)|j|jSdS(s%Retrieve the complement of the symbolN(RRTRRL(RDRyt complementi((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyRps cCs |j|S(s,Retrieve the complement of the symbol faster(R(RDRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetComplementFastwscCs(x!|D]}|j|stSqWtS(sECheck if the string contains only symbols that belong to the alphabet(RRRRp(RDtsymstrRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt isValidString{s cCs|jS(s3Retrieve the number of core symbols in the alphabet(R(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetLenscCstt|j|jdS(sERetrieve the number of core symbols plus the wildcard in the alphabeti(RRRR(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt getWildLenscCs t|jS(sERetrieve the full count of core and ambiguous symbols in the alphabet(RR(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt getFullLenscCs g|D]}|j|^qS(s,Encode the symstr as indexes of the alphabet(R(RDRRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt encodeStringscCs?|jdkr+t|jdkr+|jSdj|jSdS(sGet the alphabet nameiRN(RMRTRRR(RD((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytgetNames$cCsg}g}x|jD]}|j|jks4|r=|jn |j}|jr|jjr|j|jj|j|nxc|j D]X}|j||j||jr|jr|j|j|j|qqWqWt j dj |dj |S(sCreate a translation table that will convert sequences to the prime symbol. Optionally convert ambiguous symbols to the wildcard symbol.R( RRRRLRRR RuRRQtstringt maketransR(RDt allow_ambigtsrctdestRtdsymRy((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt translators-  c Csg}xt|jD]}|j|}|jdkr>qn|j|j}|j|krcqn|j|j|jd|j|j|jdqWxYt|jD]H}|j|}|jdkrqn|j|j|jdqWxt|jt|jD]}|j|}dj t g|j D]}|j|j ^qEdt }|j|j|jd|j||jd|jdkrxH|jD]:}|j||jd|j||jdqWqqWxt|jD]p}|j|}|jdkrxK|jD]=}|j||jd|j|j |jdqAWqqWdj |S(s.Create the text representation of the alphabets ~ s RRs = N(RRRRORTRRuRRRRRPRLR RQ( RDRR'tsym1tsym2Rytsymit comprise_strRS((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyRsP   " ;         s s csg}|jdj|tjjdkr\|jdjjjn|jdxjD]}|jdj|||jt|j t |j dkr|jdjtdj |j nt |j dkr@|jdkrxj|j}|jd jt|j qxn8|jd jtdj fd |j D|jdkrt |jdkr|jd jt|jn|jdkr|jd j|jn|jdqsW|jdj|dj |S(s-Create the XML representation of the alphabets{:s} s%{:s}{:s}ss name={:s}s colour="{:06X}"s/> s{:s} N(RuRRRRhRTRRRRLRRQRRPRORMRN(RDtpadtindentRRytcsym((RDsD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytasXMLs*%" .+%8$"(%RHRIRJRERRRRRRRRRRRRRRRRRRRRRRRR R R R R RRpRRR(((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyRsD F                        0cCst}|jddd|jddddd|jdd d dd |jd d d dd|jddddd|jdddd|jdddd|jdddd|jdddd|jdddd|jdd dd!|jd"d#dd$|jd%d&dd'|jd(d)dd*|jd+d,dd-|jd.d/dd0|jd1d2dd3|jd4d2dd3|jd5d2dd3|jt|S(6sCreate a DNA alphabetRRhRtAdenineiROtTtCtCytosineitGtGuanineitThymineitUtUracilRPtWtWeaktATtStStrongtCGtMtAminotACtKtKetotGTtRtPurinetAGtYt PyrimidinetCTtBsNot AtCGTtDsNot CtAGTtHsNot GtACTtVsNot TtACGtNsAny baseRtXt.(RYRRRR(tfactory((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytdnas,  cCst}|jddd|jddd|jddd|jd d d |jd d d|jdddd |jdddd|jdddd|jdddd|jdddd|jdddd |jd!d"dd#|jd$d%dd&|jd'd(dd)|jd*d+dd,|jd-d.dd/|jd0d1dd2|jd3d1dd2|jd4d1dd2|jt|S(5sCreate a RNA alphabetRRhRRiR!R"iR#R$iR&R'iR R%RPR(R)tAUR+R,R-R.R/R0R1R2tGUR4R5R6R7R8tCUR:sNot AtCGUR<sNot CtAGUR>sNot GtACUR@sNot URARBsAny baseRRCRD(RYRRRR(RE((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytrnas,  cCs0t}|jddd|jddd|jddd |jd d d |jd dd|jddd|jddd|jddd |jddd|jddd|jddd|jddd|jd d!d |jd"d#d|jd$d%d|jd&d'd(|jd)d*d |jd+d,d |jd-d.d|jd/d0d1|jd2d3d|jd4d5d6d7|jd8d9d6d:|jd;d<d6d=|jd>d?d6d@|jdAd?d6d@|jdBd?d6d@|jt|S(CsCreate a protein alphabettProteinRhRRtAlanineiR4tArginineiRBt AsparagineiR<s Aspartic acidiR!tCysteinetEs Glutamic acidtQt GlutamineR#tGlycineiR>t HistidineitIt IsoleucinetLtLeucineR1tLysineR.t MethioninetFt PhenylalaninetPtProlineiR+tSerineR t ThreonineR(t TryptophanR7tTyrosinei3R@tValineR:sAsparagine or Aspartic acidRPtNDRsGlutamine or Glutamic acidtQEtJsLeucine or IsoleucinetLIRCsAny amino acidtARNDCEQGHILKMFPSTWYVt*RD(RYRRRR(RE((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pytprotein's<  cCs(x!|D]}|j|kr|SqWdS(sjRetrieve a pre-defined alphabet by name. Currently, "Protein", "DNA" and "RNA" are available. Example: >>> alpha = sequence.getAlphabet('Protein') >>> alpha.getSymbols() will retrieve the 20 amino acid alphabet and output the tuple: ('A', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'K', 'L', 'M', 'N', 'P', 'Q', 'R', 'S', 'T', 'V', 'W', 'Y') N(RMRT(RMt alphabetstalphabet((sD/woldlab/castor/home/georgi/programs/meme_4.11.2/scripts/alphabet.pyt getByNameJs cCsi}x7|D]/}||kr2||cd7s6T &  #  + $ "P8r   !  :