ó Z_Sc@@sÌdZddlmZmZddlZddlmZddlm Z ddlm Z m Z m Z ddl mZdZd efd „ƒYZe dƒZd efd„ƒYZd efd„ƒYZdS(s( Arrays indexed by alphabetic strings. i(tabsolute_importtprint_functionN(tzipi(tAlphabet(tunambiguous_dna_alphabettunambiguous_rna_alphabettunambiguous_protein_alphabet(tisinttAlphabeticArraytsubmatrix_alphabett SubMatrixtMotifcB@sheZdZddgZd d d„Zd„Zd„Zd„Zd„Z d„Z d „Z d „Z RS( sIAn alphabetic array. Wraps a numpy array so that each dimension can be associated with an alphabet and indexed with characters or strings. Attributes : - alphabets -- A sequence of alphabets used to index the array - array -- The underlying array object that is indexed. Examples : >>> from corebio.seq import * >>> from corebio.matrix import AlphabeticArray >>> >>> str(protein_alphabet) 'ACDEFGHIKLMNOPQRSTUVWYBJZX*-' >>> matrix = AlphabeticArray( (protein_alphabet, protein_alphabet) ) >>> >>> # Index by character or integer: >>> matrix['A', 'C'] = 10 >>> matrix[0,1] 10 >>> >>> # Different alphabets on each dimension: >>> import numpy as na >>> a234 = na.zeros( shape = (2,3,4) ) >>> alpha = ( "AB", "ABC", "ABCD") >>> aa = AlphabeticArray(alpha,a234) >>> aa['A', 'B', 'C'] = 22 >>> >>> # String indices are converted to integer index arrays: ... >>> aa['A', 'B', 'ABCD'] array([ 0, 0, 22, 0]) Authors: o GEC 2005, JXG 2006 t alphabetstarrayc C@sªdtfd„ƒY}g}g}xª|D]¢}t|tƒrMt|ƒ}n|dkry|jdƒ|j|ƒƒq)t|tƒr«|jt|ƒƒ|j|ƒq)|jt|ƒƒ|jdƒq)Wt|ƒ}|dkrt j d|d|ƒ}nŒt j |d|ƒ}|j }t|ƒt|ƒkrGt dƒ‚nxDt||ƒD]3\} } | dk rW| | krWt dƒ‚qWqWW||_t|ƒ|_dS(sé Args: - alphabets -- a list of alphabets (as string or Alphabet objects) to be used to convert strings into indices. The lengths of the alphabets match the shape of the indexed array. Alternatively, an integer or None in the list indicate a non-alphabetic dimension. If None the dimension length is taken from values argument. - values -- An array of values to be indexed. If None a new array is created. If this argument is not a numpy array then the alphabet list must be explicit (cannot contain None.) - dtype -- An optional numpy type code. t NullAlphabetcB@seZd„Zd„ZRS(cS@stdƒ‚dS(Ns(This dimension does not have an alphabet(t IndexError(tselftkey((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pytordlscS@stdƒ‚dS(Ns(This dimension does not have an alphabet(R(RR((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pytordsns(t__name__t __module__RR(((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyRks tshapetdtypes$The values array is the wrong shape.N(tobjectt isinstancetstrRtNonetappendtlentintttupletnatzerostasarrayRt ValueErrorRR R ( RR tvaluesRRtalphaRtatvshapets1ts2((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyt__init__Xs4       cC@s|jj|j|ƒƒS(N(R t __getitem__t_ordkey(RR((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyR+scC@s |jj|j|ƒ|ƒdS(N(R t __setitem__R,(RRtvalue((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyR-’sc@sn‡fd†‰t|tƒrVtgt||jƒD]\}}ˆ||ƒ^q4ƒSˆ||jdƒSdS(s¤Convert string indices into integers. Handles characters, strings slices with strings, and tuples of the same. Anything else is unchanged. c@sÙ|dkrdSt|tƒs.t|tƒr…t|ƒ}t|ƒdkrY|j|ƒSt|ƒdkrodStj|j|ƒƒSt|t ƒrш|j |ƒ}ˆ|j |ƒ}|j }t |||ƒS|SdS(Nii( RRRRRRR R"Rtslicetstarttstoptstep(RR%R0R1R2(tnorm(s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyR3›s    iN(RRRR (RRtkR&((R3s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyR,•s8cC@sÄ|j|ƒ}g}x–t|ƒD]ˆ\}}|dkrVtd|jj|ƒ}ntj|ƒ}x5tt|ƒ|dƒD]}|dtj f}q€W|j |ƒq"W|jj t |ƒƒS(sµ Return an array of shape (len(key1), len(key2), ...) whose values are indexed by keys. a.outerindex( (I,J,K) )[i,j,k] == a.array[I[i],J[j],K[k]] ii.N( R,t enumerateRtrangeR RR R"RtnewaxisRR+R(Rtkeyst outerkeystiR4tj((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pytindex°s  !cC@s|j|ƒ}t||ƒS(sÌCreate a new AlphabeticArray with the given alphabets. The new alphabet must be a subset of the current alphabet. Useful for extracting a submatrix or for permuting the alphabet. (R<R(Rt new_alphabetst new_array((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pytreindexÅscC@s9ytj||ƒSWntk r4t|j|ƒSXdS(N(Rt __getattr__tAttributeErrortgetattrR (Rtname((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyR@Ñs cC@s?ytj|||ƒSWn!tk r:t|j||ƒSXdS(N(Rt __setattr__RAtsetattrR (RRCR.((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyRD×s N( RRt__doc__t __slots__RR*R+R-R,R<R?R@RD(((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyR,s& 7     tARNDCQEGHILKMFPSTWYVBZXcB@s\eZdZdddddgZd d d d d d„Zd„Zed ej d„ƒZ RS( spA two dimensional array indexed by an Alphabet. Used to hold substitution matrices and similar information. Various standard substitution matrices are available from the data package >>> from corebio import data >>> mat = SubMatrix.read(data.data_stream('blosum100')) Attr: - alphabet -- An Alphabet - array -- A numpy array - name -- The name of this matrix (if any) as a string. - description -- The description, if any. - scale -- The scale constant of a log-odds matrix, if known. Authors: o GEC 2005, JXG 2006 talphabetR RCt descriptiontscalecC@sJtj|||f||ƒt|ƒ|_||_||_||_dS(N(RR*RRIRCRJRK(RRIR ttypeofRCRJRK((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyR*s   cC@stj|||fƒS(N(RR?(RRI((s8/home/psgendb/BIRCHDEV/pkg/weblogo-3.4/corebio/matrix.pyR? sc C@s@|d krt}nt|ƒ}tj||f|ƒ}d}xit|ƒD][\}}|jƒsL|ddksL|ddkrŠqLn|jƒ}|d|dkr°qLn|djƒrø|d||krøt d||d||fƒ‚n|djƒr|d}nt|ƒdkr4|d }nt|ƒ|krYt d|ƒ‚nx1t d|ƒD] } t || ƒ||| fs ¶ x