3 h}\XH@s|dZddlZddlmZddlmZmZmZddlm Z m Z dZ Gd dde Z ed ZGd d d e ZGdd d e ZdS)z' Arrays indexed by alphabetic strings. N)Alphabet)unambiguous_dna_alphabetunambiguous_rna_alphabetunambiguous_protein_alphabet)isintischarAlphabeticArraysubmatrix_alphabet SubMatrixMotifc@sZeZdZdZddgZdddZddZd d Zd d Zd dZ ddZ ddZ ddZ dS)r a'An 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 weblogo.seq import * >>> from weblogo.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 = 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 alphabetsarrayNc Cs$Gdddt}g}g}xz|D]r}t|tr4t|}|dkrT|jd|j|qt|trx|jt||j|q|jt||jdqWt|}|dkrtj ||d}n\tj ||d}|j }t|t|krt dx.t ||D] \} } | dk r| | krt dqW||_t||_dS)a Args: - alphabets -- a list of alphabets (as string or Alphabet objects) to be used to convert strings into indices. The lengths of the alphabets match the shape of the indexed array. Alternatively, an integer or None in the list indicate a non-alphabetic dimension. If None the dimension length is taken from values argument. - values -- An array of values to be indexed. If None a new array is created. If this argument is not a numpy array then the alphabet list must be explicit (cannot contain None.) - dtype -- An optional numpy type code. c@seZdZddZddZdS)z.AlphabeticArray.__init__..NullAlphabetcSs tddS)Nz(This dimension does not have an alphabet) IndexError)selfkeyr@/home/birch/.local/lib/python3.6/site-packages/weblogo/matrix.pyordhsz2AlphabeticArray.__init__..NullAlphabet.ordcSs tddS)Nz(This dimension does not have an alphabet)r)rrrrrordsksz3AlphabeticArray.__init__..NullAlphabet.ordsN)__name__ __module__ __qualname__rrrrrr NullAlphabetgsrN)shapedtype)rz$The values array is the wrong shape.)object isinstancestrrappendleninttuplenpzerosasarrayr ValueErrorziprr ) rr valuesrralpharaZvshapes1s2rrr__init__Ts4      zAlphabeticArray.__init__cCs|jj|j|S)N)r __getitem___ordkey)rrrrrr.szAlphabeticArray.__getitem__cCs|jj|j||dS)N)r __setitem__r/)rrvaluerrrr0szAlphabeticArray.__setitem__csHfddt|tr4tfddt||jDS||jdSdS)zConvert string indices into integers. Handles characters, strings slices with strings, and tuples of the same. Anything else is unchanged. cs|dkr dSt|ts t|tr^t|}t|dkr>|j|St|dkrNdStj|j|St|tr|j |}|j |}|j }t|||S|SdS)Nrr) rrrr rr#r%rslicestartstopstep)rr)r3r4r5)normrrr6s       z%AlphabeticArray._ordkey..normcsg|]\}}||qSrr).0kr*)r6rr sz+AlphabeticArray._ordkey..rN)rr"r'r )rrr)r6rr/s  zAlphabeticArray._ordkeycCs|j|}g}xnt|D]b\}}|dkr:td|jj|}tj|}x*tt||dD]}|dtjf}qZW|j |qW|jj t |S)z 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]] Nrr.) r/ enumeraterangerrr#r%r Znewaxisrr.r")rkeysZ outerkeysir8jrrrindexs  zAlphabeticArray.indexcCs|j|}t||S)zCreate a new AlphabeticArray with the given alphabet. The new alphabet must be a subset of the current alphabet. Useful for extracting a submatrix or for permuting the alphabet. )r?r )rZ new_alphabetZ new_arrayrrrreindexs zAlphabeticArray.reindexc Cs.y tj||Stk r(t|j|SXdS)N)r __getattr__AttributeErrorgetattrr)rnamerrrrAs zAlphabeticArray.__getattr__c Cs2ytj|||Stk r,t|j||SXdS)N)r __setattr__rBsetattrr)rrDr1rrrrEszAlphabeticArray.__setattr__)NN) rrr__doc__ __slots__r-r.r0r/r?r@rArErrrrr (s& 8 ZARNDCQEGHILKMFPSTWYVBZXc@sDeZdZdZdddddgZddd Zd d Zedej fd d Z dS)r aVA 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 weblogo 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 alphabetrrD descriptionscaleNcCs4tj|||f||t||_||_||_||_dS)N)r r-rrIrDrJrK)rrIrtypeofrDrJrKrrrr-s  zSubMatrix.__init__cCstj|||fS)N)r r@)rrIrrrr@szSubMatrix.reindexc Cs|dkr t}t|}tj||f|}d}xt|D]\}}|js4|ddks4|ddkr`q4|j}|d|dkrzq4|djr|d||krtdj ||d|||djr|dd}t|dkr|dd}t||krtd j |x(t d|D]} t || ||| f<qW|d7}||kr4Pq4Wtd xTt d|D]F}x>t d|D]0} ||| f|| |fkr^td j || q^WqNWt ||S) a] Parse and return a substitution matrix Arguments: - fin -- matrix file - alphabet -- The set of substitution characters. Default: '' - typeof -- A numpy type or typecode. Returns: - A numpy matrix of substitution scores Raises: - ValueError on unreadable input Nr#*rz(Incompatible alphabet: line {} : {} {}: z$SubMatrix matrix parseerror: line {}z Premature EOFz+Substitution matrix is asymmetric! ({}, {})) r r r#r$r:isspacesplitisalphar&formatr;floatr ) finrIrLLmatrixr=Zlinenumlinecellsr>rrrreadsB      zSubMatrix.read)NNNNN) rrrrGrHr-r@ staticmethodr#float64r[rrrrr s c@s\eZdZdZddgZdddZeddZd d Zd d Z d dZ ddZ e dddZ dS)r aVA two dimensional array where the second dimension is indexed by an Alphabet. Used to represent sequence motifs and similar information. Attr: - alphabet -- An Alphabet - array -- A numpy array - name -- The name of this motif (if any) as a string. - description -- The description, if any. ZXXz//NcCs*tj|d|f||||_||_||_dS)N)r r-rDrJrK)rrIrrrDrJrKrrrr-hszMotif.__init__cCs |jdS)Nr)r )rrrrrIoszMotif.alphabetcCst|tj|d|fS)N)r r r@)rrIrrrr@ssz Motif.reindexcCs|jddd|_dS)zReverse sequence dataNr)r)rrrrreverse{sz Motif.reversecCsRddlm}m}|j}||||j}d|f|_|j|}d|f|_|j|_dS)z!Complement nucleic acid sequence.r)SeqrN)Z weblogo.seqr`rrI complementr r@r)rr`rrIZcomplement_alphabetmrrrras   zMotif.complementcCs|j|jdS)zkComplements and reverses nucleic acid sequence (i.e. the other strand of a DNA sequence.) N)r_ra)rrrrreverse_complementszMotif.reverse_complementcCsvg}d}xf|D]^}|js|ddkr(q|j}|ddksH|ddkrLd}|r|d|jkrbPq|j|qWt|dkrtd|jd}t|}t|} t|d} |ddkp|ddkp|| d kp|| dkstd t|d} x8td t|D]&} | t|| krtd j| qW|ddksD|ddkrN|jdd} x4|D],} t | sttd j| t | sXd} qXW| rdnd}|rxt |D]N\} }t |d r|ddd krtdj| |jddj |}qWnfg}xVt |D]J\} }t |d r@|ddd kr@tdj| |j|jdqWdj |}t |}|rt |}|j|s|}n6tttf}x |D]}|j|r|}PqW|s|}t|dt|d krx|D]}|jqWt|} t|d} tj| | ftjd}x>t| D]2}x*t| D]}t||||||f<q0Wq"W| rf|jt||j|S)z Parse a TRANSFAC-format PWM from a file. Returns a Motif object, representing the provided PWM along with an inferred or provided alphabet. FrrMZPOZP0Tz Vacuous file.rzMissing header line!zInconsistant length, row: {}zEExpected a single character per header item, but got "{}" as one itemPz*Expected position as first item on line {})r)rQrR_TRANSFAC_DELIM_LINESrr r&popr;rTrrr:joinrZ alphabeticrrrr#r$r]rUZ transposer r@)clsrVrIitemsr3rYstuffheaderZhcolsZrowscolsr=Zposition_headerhZalphabet_headerrZdefacto_alphabetr*r rXcrrr read_transfacs          " "      $zMotif.read_transfac)NNNNN)N)rrrrGrgr-propertyrIr@r_rarc classmethodrrrrrrr Ys    )r r r r )rGnumpyr#seqrrrrutilsrr__all__rr r r r rrrrs 7x