tf"ddZddlZddlZejddZejddZdZiddd d d d d d d dd d dd dd ddddddddddddddddd d!d"d#d$d%d&Zidd d d d dd dddddddddddddddd!dd"d'd#d'd$d(d%d(Zd)Z d*Z d+Z d,Z d-Z y).z>Tools for working with files in the samtools pileup -c format.NPileupSubstitution) chromosomeposreference_basegenotypeconsensus_quality snp_qualitymapping_qualitycoverage read_basesbase_qualities PileupIndel) rrrrrr r r first_allele second_allele reads_first reads_second reads_diffc #Ktdttttttttf }tdtttttttttttf }|D]z}|ddj}|ddk(r/ tt ||Dcgc] \}}||c}}M tt ||Dcgc] \}}||c}}|ycc}}w#t $rt jd|zwxYwcc}}w#t $rt jd|zwxYww)a/iterate over ``samtools pileup -c`` formatted file. *infile* can be any iterator over a lines. The function yields named tuples of the type :class:`pysam.Pileup.PileupSubstitution` or :class:`pysam.Pileup.PileupIndel`. .. note:: The parser converts to 0-based coordinates ct|dz SNintxs V/var/www/html/software/conda/envs/higlass/lib/python3.12/site-packages/pysam/Pileup.pyziterate..0Q!ct|dz Srrrs rrziterate..2rrN*zparsing error in line: `%s`) strrsplitrzip TypeErrorpysam SamtoolsErrorr)infile conv_subst conv_indellinedrys riterater0#s*+SsCc35J+S#ssCc3S:J P "IOO  Q43; P!SQ5G#HTQAaD#HII P(C A.getPrefixqsP BR !q A!u1~"1v  "1v rctt|t|}|d|dk7rytd| dz dD]}||||k7s||dzdcS|| dS)z&get common sufix of strings s1 and s2.r!rNr]ras r getSuffixz0translateIndelGenotypeFromVCF..getSuffixysu BR ! b6RV rA262& "A!u1~!a%&z! "1"#wrc||k(ryt|t|kDr|j|rd|t|dzt|dz fS|j|rd|dt| zdfS||}||}t|t|zt|z }|dkr td|t|t||z zt|dz fSt|t|kr|j|rd|t|dzt|dz fS|j|rd|dt|zdfS||}||}t|t|zt|z }|dkr td|t|t||z zt|fSJd)N)r#rz-%srr!rz+%szsnp?)r_ startswithendswith ValueError)variantrefprefixsuffixsharedreris r getGenotypez2translateIndelGenotypeFromVCF..getGenotypes c> s8c'l "~~g&s3w<=113wS\M22B66"30"30Vs6{2S\AA:$,&s3v;Vv1E/FGGVWXXX XG $!!#&ws3xy113s8a<??!!#&wyC11144"30"30Vs6{2SX=A:$,&ws6{S[65I3JKKSQW[XX f 1rTr#Frzmultiple offsets for indelr/)rmappendr_setjoin) vcf_genotypesrors genotypesoffsetsis_errorrngoffsetreris @@rtranslateIndelGenotypeFromVCFr~ms#RRwIH  #GS1IAv  8 NN6 " l s7|  !?#?? ! QZF#I f #   s B B$#B$c|j}|j}|j}|g|jz}||}|d}t |dkDrt dt |z|d}|ddk(ry|Dcgc]}|dk7s |t|}}|jddgdx} } |jjd dgd} |jd d} t |dkDs/t|jDcgc] }t |c}dkDr5t||\} }t|||zd | | | | | | d t | zddd Std j|} d }d }t|||| | | | | || Scc}wcc}w)z$convert vcf record to pileup record.rFrz%only single genotype per position, %sr.NrtGQMQDPr#Fc#Ktj}|j|||jvr t d|j D]}t ||}|s|yw)aiterate over a vcf-formatted file. *infile* can be any iterator over a lines. The function yields named tuples of the type :class:`pysam.Pileup.PileupSubstitution` or :class:`pysam.Pileup.PileupIndel`. Positions without a snp will be skipped. This method is wasteful and written to support same legacy code that expects samtools pileup output. Better use the vcf parser directly. zsample %s not vcf fileN)r(VCFconnect getsamplesKeyErrorfetchr)r*rrrowresults riterate_from_vcfrsc" ))+CKK S^^%%/00yy{C( Ls A$A.'A.)__doc__ collectionsr( namedtuplerrr0rSrYrWrZr~rrrrrsD +[++,@ -B C%k$$] &78 "P@  3 S "%s # S  %)3 04c  # S  # S   #  S   #  S  S S         D    D  D  D  D  D  )!Vr72tr