Xf5]dZddlZddlZddlZddlmZddlmZGdde Z Gdd e Z Gd d e Z Gd d e Z dS)zSupport for various forms of sequence motif matrices. Implementation of frequency (count) matrices, position-weight matrices, and position-specific scoring matrices. N)Seq)_pwmceZdZdZdZdZdZedZedZ edZ d d Z ed Z d Z dS)GenericPositionMatrixz9Base class for the support of position matrix operations.cd|_|D]f}|jt|||_n-|jt||krtdd||D||<g||_dS)Initialize the class.Nzdata has inconsistent lengthsc,g|]}t|S)float).0_s 1lib/python3.11/site-packages/Bio/motifs/matrix.py z2GenericPositionMatrix.__init__..$s===E!HH===)lengthlen Exceptionalphabet)selfrvaluesletters r__init__zGenericPositionMatrix.__init__s  > >F{"!&.11 F6N 3 333 ?@@@==fVn===DLL  rc@dt|jD}dd|z}|g}|jD]D}d||D}d|zd|z}||Ed|dz}|S)zPReturn a string containing nucleotides and counts of the alphabet in the Matrix.cg|]}d|zS)z%6dr )r is rrz1GenericPositionMatrix.__str__..)s777q777rz  cg|]}d|zS)z%6.2fr )r values rrz1GenericPositionMatrix.__str__..-s???Wu_???rz%c:  )rangerjoinrappend)rwordslinelinesrtexts r__str__zGenericPositionMatrix.__str__'s77E$+$6$6777sxx&m  F??$v,???EF?SXXe__4D LL    yy$& rct|trt|dkr|\}}t|trM|tj\}}}t |||}fd|D}d} nt|tjrj|} d} ntt|trfd|D}d} nNt|tr't|dkr|} d} n!t|td|zt|tr2|j \} } } t | | | }d}n1t|tjr|}d}ntd|z| dkr'|dkr!t | |S| dkr<|dkr6t | tfd|DS| dkr3|dkr-i}|D]&} t | ||| <'|Si}|D].} t | fd|D|| </t|jkrj|S|St|dkr |d}ntd t|trM|tj\}}}t |||}fd |D}d}nt|tjrj|}d}ntt|trfd |D}d}nNt|tr't|dkr|}d}n!t|td|z|dkrt |S|dkr'i}|D] }t |||<!|St!d ) z(Return the position matrix of index key.c*g|]}j|Sr rr rrs rrz5GenericPositionMatrix.__getitem__..;s CCCQ a 0CCCrrc*g|]}j|Sr r,r-s rrz5GenericPositionMatrix.__getitem__..As ???Q a 0???rzCannot understand key %sc3(K|] }|V dSNr )r index2rs r z4GenericPositionMatrix.__getitem__..Xs' G GF G G G G G Grc g|] }| Sr r )r rrs rrz5GenericPositionMatrix.__getitem__..bs%B%B%BAfQi%B%B%Brrz"keys should be 1- or 2-dimensionalc*g|]}j|Sr r,r-s rrz5GenericPositionMatrix.__getitem__..ns 999At}Q'999rc*g|]}j|Sr r,r-s rrz5GenericPositionMatrix.__getitem__..ts 555At}Q'555rzShould not get here) isinstancetuplersliceindicesrr!numbersIntegralstrKeyErrorrdict __getitem__sorted __class__ RuntimeError)rkeykey1key2start1stop1stride1indices1letters1dim1letter1start2stop2stride2indices2dim2r1dstartstopstrider9lettersdimrrs` @rr?z!GenericPositionMatrix.__getitem__3s c5 ! !5 E3xx1}} ddE**F-1\\#dm:L:L-M-M*FE7$VUG<.getsJ'**rT)rCreversec,g|]}|Sr r )r crrs rrz>GenericPositionMatrix.degenerate_consensus..s!666Qd1gaj666rrrNr*r)r!rr@sumr"rr) rdegenerate_nucleotider]r nucleotidescountsrCrrs ` @rdegenerate_consensusz*GenericPositionMatrix.degenerate_consensuss! ! "t{## # #A + + + + + +!3===K66666+666Fay3vabbz??**vay1vay=/H/H!!nS__$q3v;;66ggf[!_5566aggf[!_5566.223<>V Q/Hfa( )%rcHi}|jdkr)|dddd|d<|dddd|d<n(|dddd|d<|dddd|d<|dddd|d<|dddd|d<|j}|||S) zCompute reverse complement.ACGUUNrfrirgrh)rrA)rrrs rreverse_complementz(GenericPositionMatrix.reverse_complement%s =F " "s)DDbD/F3Ks)DDbD/F3KKs)DDbD/F3Ks)DDbD/F3K3i"os 3i"os =~~h///r)NNrN)__name__ __module__ __qualname____doc__rr(r?propertyrardrrrrr rrrrsCC ! ! !   S6S6S6j  X   X ++X+\MQDDDDL   X   0 0 0 0 0rrceZdZdZddZdS)FrequencyPositionMatrixzGClass for the support of frequency calculations on the Position Matrix.Nci}||jD]}dg|jz||<nft|tr,|jD]#}t ||g|jz||<$n%|jD]}t |g|jz||<t |jD].}|jD]$}|||xx|||z cc<%/t |j|S)aCreate and return a position-weight matrix by normalizing the counts matrix. If pseudocounts is None (default), no pseudocounts are added to the counts. If pseudocounts is a number, it is added to the counts before calculating the position-weight matrix. Alternatively, the pseudocounts can be a dictionary with a key for each letter in the alphabet associated with the motif. Nr)rrr6r>r r!PositionWeightMatrix)r pseudocountsrrrs r normalizez!FrequencyPositionMatrix.normalize7s'  - 5 5"%!4v 5  d + + E- M M"' V(<"="=!>!Lv M- E E"' "5"5!6!Dvt{## 5 5A- 5 5vq!!!T&\!_4!!!! 5$DM6:::rr0)rrrrrr rrrr4s.QQ;;;;;;rrc eZdZdZdZddZdS)rzDClass for the support of weight calculations on the Position Matrix.ct||tjD]9t fd|D}|D]}|xx|zcc<:|D]}t ||<dS)r c34K|]}|VdSr0r r rrrs rr2z0PositionWeightMatrix.__init__..[s+??FV Q??????rN)rrr!rrr7)rrrrrrs` @rrzPositionWeightMatrix.__init__Ws&&tXv>>>t{## ) )A?????h?????E" ) )V Q5( ) / /F f..DLL / /rNcli}|j}|!t|jd}nt|}t|}|D]}||xx|zcc<g||<t |jD]}|D]}||}|dkr;|||}|dkrtj||z d} n;tj } n-|||}|dkr tj } n tj } || | t||} | S)a/Return the Position-Specific Scoring Matrix. The Position-Specific Scoring Matrix (PSSM) contains the log-odds scores computed from the probability matrix and the background probabilities. If the background is None, a uniform background distribution is assumed. N?rr*) rr>fromkeysrrr!rr[logr\nanr#PositionSpecificScoringMatrix) r backgroundrrrrrbplogoddspssms rlog_oddszPositionWeightMatrix.log_oddsasX=  t}c::JJj))JJ%%''((  F v   % '   F6NNt{## / /A" / /v&q55V QA1uu"&(1q5!"4"4#'8)V QA1uu"&("&(v%%g.... /-Xv>> rr0)rrrrrrr rrrrTs=NN///######rrceZdZdZdZddZedZedZed Z dd Z dd Z d Z dZ ddZd S)rzGClass for the support of Position Specific Scoring Matrix calculations.ctjgdkrtdjz t|}nr#t$rL t|d}n7#t$rtddt $rtddwxYwYnt $rtddwxYwt|}j}tj ||z dztj }tj fdt|Dt}tj|||t|dkr|d S|S) agReturn the PWM score for a given sequence for all positions. Notes: - the sequence can only be a DNA sequence - the search is performed only on one strand - if the sequence and the motif have the same length, a single number is returned - otherwise, the result is a one-dimensional numpy array )rfrgrhriz6PSSM has wrong alphabet: %s - Use only with DNA motifsASCIIzBsequence should be a Seq, MutableSeq, string, or bytes-like objectNz-sequence should contain ASCII characters onlyrc0g|]fddDS)c,g|]}|Sr r rs rrzFPositionSpecificScoringMatrix.calculate...s! 3 3 3&d6l1o 3 3 3rrr r-s @rrz;PositionSpecificScoringMatrix.calculate..s2 F F F 3 3 3 3 3F 3 3 3 F F Frr)r@rrbytes TypeErrorUnicodeEncodeErrorrrrnpemptyfloat32arrayr!r r calculate)rr]nmscoresrs` rrz'PositionSpecificScoringMatrix.calculates $- $8$8$8 8 8H4=X  XHH     733    X&    C     T   MM K!a%!)RZ00( F F F FU1XX F F F   x&111 v;;!  !9 Ms)A B3A B3 4BB3B3rT@Bc#PK|}t|}|j}tjd||}|r|}|D]I} || | |z|zdz } || } | |k} tj| d| z} | | } |rA|| }||k}tj|d| z}||}n6tjdt}tjdt}tj | ||z }tj | |}tj tj | |}||}||}t||Ed{VKdS)zFind hits with PWM score above given threshold. A generator function, returning found hits in the given sequence with the pwm score higher than the threshold. rr)dtypeN) upperrrrarangerrwhererintr#argsortzip)rr] thresholdboth chunksizeseq_lenmotif_l chunk_startsrc chunk_startsubseq pos_scorespos_ind pos_positions neg_scoresneg_ind neg_positionschunk_positions chunk_scoresorders rsearchz$PositionSpecificScoringMatrix.searchs >>##h--+yGY77  +((**B' : :KkK),Cg,MPQ,QQRF//J I-GHW--a0;>M#G,J 6\\&11 $ 1 " 1 1! 4{ B '0 "!C 8 8 8 Xq555  i }w7NOOO9Z<.,FFFfh/FFFFFFr)rr!rmaxrscorerVrs` @rrz!PositionSpecificScoringMatrix.max^ -dk** G GH SFFFFFgFFFFF FEE rcd}j}tjD]!|tfd|Dz }"|S)zsMinimal possible score for this motif. returns the score computed for the anticonsensus sequence. rc34K|]}|VdSr0r rs rr2z4PositionSpecificScoringMatrix.min..rr)rr!rminrs` @rrz!PositionSpecificScoringMatrix.minrrc td)zCompute the GC-ratio.z,Cannot compute the %GC composition of a PSSM)r)rs rrz(PositionSpecificScoringMatrix.gc_contentsFGGGrNc|!t|jd}nt|}t|}|jD]}||xx|zcc<d}t |jD]n}|jD]d}|||f}tj|r!tj |r|dkr<||}|tj d|z}|||zz }eo|S)z.Return expected value of the score of a motif.Nrrrr*) r>rrrrr!rr[isnanisinfpow) rrrrsxrrrrs rmeanz"PositionSpecificScoringMatrix.means  t}c::JJj))JJ%%''((m ( (F v   % '     t{## " "A- " "vqy/:g&&:g&&7Q;;v&G,,,a'k! " rcf|!t|jd}nt|}t|}|jD]}||xx|zcc<d}t |jD]}d}d}|jD]o}|||f}tj|r!tj |r|dkr<||} | tj d|z} || |zz }|| |z|zz }p|||zz}||z }t|d}tj |S)z2Return standard deviation of the score of a motif.Nrrrr*) r>rrrrr!rr[rrrrsqrt) rrrrvariancerrsxxrrrs rstdz!PositionSpecificScoringMatrix.stdsg  t}c::JJj))JJ%%''((m ( (F v   % '    t{##  ABC- - -vqy/:g&&:g&&7Q;;v&G,,,a'k!q7{W,, 27NC OHHx##y"""rc|j|jkrtdd}t|j dz|jD]A}|dkr||| }n|||}||kr|}| }Bd|z |fS)zReturn the similarity score based on pearson correlation for the given motif against self. We use the Pearson's correlation of the respective probabilities. z.Cannot compare motifs with different alphabetsrr)rrr!rdist_pearson_at)rothermax_poffsetrmax_os r dist_pearsonz*PositionSpecificScoringMatrix.dist_pearson/s =EN * *MNN NT[L1,el;;  Fzz((88))$77qyy5y%rc j}d}d}d}d}d}tjjzt|z} t t jz jD]fd|D} fd|D} |t | z }|t | z }|t d| Dz }|t dt| | Dz }|t d| Dz }|| z}|| z}|| z}|| z}|| z}|||zz } tj |||zz |||zz z} | | z S)zMReturn the similarity score based on pearson correlation at the given offset.rc*g|]}|zfSr r )r rr posrs rrzAPositionSpecificScoringMatrix.dist_pearson_at..Ls'CCC$vsV|+,CCCrc$g|] }|f Sr r )r rrrs rrzAPositionSpecificScoringMatrix.dist_pearson_at..Ms";;;% $;;;rc3 K|] }||zV dSr0r )r xs rr2z@PositionSpecificScoringMatrix.dist_pearson_at..P&))q1u))))))rc3&K|] \}}||zV dSr0r )r rys rr2z@PositionSpecificScoringMatrix.dist_pearson_at..Qs*55Aq1u555555rc3 K|] }||zV dSr0r )r rs rr2z@PositionSpecificScoringMatrix.dist_pearson_at..Rrr) rrrrr!rrrr[r)rrr rVrsyrsxysyynormxiyi numerator denominatorrs``` @rrz-PositionSpecificScoringMatrix.dist_pearson_atBs-  4; 566WET[615<@@AA * *CCCCCCC7CCCB;;;;;7;;;B #b''MB #b''MB 3))b))))) )C 355R55555 5C 3))b))))) )CC d  d  t  t  t "r'M irBw3b= ABB ;&&rcddlm}|!t|jd}nt|}t |}|jD]}||xx|zcc<||||S)z@Calculate the distribution of the scores at the given precision.r)ScoreDistributionNr) precisionrr) thresholdsr!r>rrrr)rrr"r!rrs r distributionz*PositionSpecificScoringMatrix.distribution\s111111  t}c::JJj))JJ%%''((m ( (F v   % '      94JWWWWr)rTrr0)Nr)rrrrrrrrrrrrr rr$r rrrrsQQ333j::::B  X   X HHXH,####8   &'''4 X X X X X Xrr)rr[r:numpyrBio.SeqrrYrr>rrrrr rrr's'  Y0Y0Y0Y0Y0DY0Y0Y0x;;;;;3;;;@000000000f`X`X`X`X`X$9`X`X`X`X`Xr