a d6:@sddlZddlZddlZddlZddlZddlZddlZddlZddlZddZ ddZ ddZ dd d Z d d Z d dZddZddZdddZdddZdS)NcCs"tttdtt|dS)z@Transforming error rate to ASCII character using the Phred scalei!)chrintroundmathZlog10abs)Zpvalr0lib/python3.9/site-packages/mapdamage/rescale.pyphred_pval_to_char sr cCsdtt|td dS)zATransforming ASCII character in the Phred scale to the error rate r)floatord)Zchrrr phred_char_to_pvalsrc CsDtj|d}tj|s,td|dzt|}t|}i}|D]V}|d|vrztd||j ||dfqJt |dt |dd|t |d<qJt | D] }|| ks||kr||q|Wd WS1s0YWnHtjy>} z,td tj|d|j | fWYd } ~ n d } ~ 00d S) z Reads the damage probability correction file, returns a dictionary with this structure position (one based) - CT - probability - GA - probability zStats_out_MCMC_correct_prob.csvzRMissing file, the file Stats_out_MCMC_correct_prob.csv should be in the folder z2 Did you run the MCMC estimates of the parameters?ZPositionz;This file has multiple position definitions %s, line %d: %sC.TG.A)rrNzFile %s, line %d: %s)ospathjoinisfilesysexitopencsvZ DictReaderZline_numr rlistkeyspopError) folderrescale_length_5prescale_length_3p full_pathfiZ fi_handle corr_problinekeyerrr get_corr_probs*    ( (r&bothc Cs|dkr t|dkr"|dkr"d}n|dkr8|dkr8d}ndS||d}||vr^|||}nd}||vrx|||} nd} |d kr|S|d kr| S|d kr|t|kr|S| Sntd d S)a The position specific damaging correction, using the input corr_prob dictionary holding the damage correcting values nt_seq nucleotide in the sequence nt_ref nucleotide in the reference pos relative position from the 5' end length length of the sequence direction which end to consider the rescaling returns the correction probability for this particular set rTCrAGrforwardZbackwardr'z-Abnormal direction in the rescaling procedureN) SystemErrorr SystemExit) r"nt_seqnt_refposlength directionsubsZback_posZp5_corrZp3_corrrrr corr_this_base3s.   r6cCsptttttdddgd}||||||||ddddddddddd}|S)zJInitialize a substitution table, to track the expected substitution countsr) CT-before TC-before GA-before AG-beforeCT-afterTC-afterGA-afterAG-afterr*r)r+r(CT-pvalsCT-pvals_beforeTC-pvalsGA-pvalsGA-pvals_beforeAG-pvals)dictrziprangecopy)Zper_qualr5rrr initialize_subsis*"rKcCsl|dkr>|dkr>d}|d|7<|ddt|7<n|dkr||dkr|d }|d |7<|d dt|7<n|dkr|dkrd }|d dt|7<||krtdnB|dkr|dkrd}|ddt|7<||krtdnd}|dkrN||dtt|dd7<||dtt|dd7<|dvrh||d7<dS)zU record the expected substitution change, prob_corr is the excact version for nt_qualr(r)ZCTrArBr,r*r+ZGArDrEZTCrCz=Internal error: rescaling qualities for the wrong transitionsZAGrFZNNz-beforerz-after)r*r)r+r(N)rr.rr )r5r0r1nt_qualZ nt_newqualZ prob_corrZsub_typerrr record_subss0   $$ rMcCspdD]f}dD]\}||D]N}||kr|dt|}||vrV|||||7<q|||||<qq qdS)z;Calculates summary statistics for the substition table subs)r9r:r;r<r=r>r?r@)rr (z-QN)str)r5ilvZqvr$rrr qual_summary_subss rTcCstd|ddkrJtdt|d|ddt|d|dntd|d dkrtd t|d |d dt|d |d ntd |d dkrtdt|d|d dt|d|d ntd|ddkr*tdt|d|ddt|d|dntdtdtdt|ddt|dtdt|ddt|dtdt|ddt|dtd t|d!dt|d"td#t|d$dt|d%td&t|d'dt|d(td)t|d*dt|d+td,t|d-dt|d.td/t|d0dt|d1td2t|d3dt|d4d5S)6zPrint the substition tablezh The expected substition frequencies before and after scaling using the scaled qualities as probalities:r)rz CT rBz rAz CT NA NAr(z TC rCz TC NA NAr+z GA rErDz GA NA NAr*z AG rFz AG NA NAz) Quality metrics before and after scalingz CT-Q0 z CT-before-Q0z CT-after-Q0z CT-Q10 z CT-before-Q10z CT-after-Q10z CT-Q20 z CT-before-Q20z CT-after-Q20z CT-Q30 z CT-before-Q30z CT-after-Q30z CT-Q40 z CT-before-Q40z CT-after-Q40z GA-Q0 z GA-before-Q0z GA-after-Q0z GA-Q10 z GA-before-Q10z GA-after-Q10z GA-Q20 z GA-before-Q20z GA-after-Q20z GA-Q30 z GA-before-Q30z GA-after-Q30z GA-Q40 z GA-before-Q40z GA-after-Q40N)printrQ)r5rrr print_subss0 6 6 66$$$$$$$$$rVFc Csd|stt}|j}tj|} ||j} | | t | t |  } tj |j||jd| \} } } t|}t| }|jrtj| } tj| } | ddd} dg|}d}d}tt|| | | D]\}}}}|dkr|dks|dkr>|d kr>d t||||d ||d }d t|}||}td |}|d |7}nd t|}|}||kr|||<t||||||||d kr|d 7}q|s|d |jqqd|}|jr|ddd}|jdddkr|jd|jdd |}|jdddkr(||j|jdd  d}||_td|}|drRt d||!d|d|S)a bam a pysam bam object read a pysam read object ref a pysam fasta ref file reflengths a dictionary holding the length of the references subs a dictionary holding the corrected number of substition before and after scaling corr_prob dictionary from get_corr_prob returns a read with rescaled quality score Iterates through the read and reference, rescales the quality according to corr_prob iNrr8r(r)r*r+r,)r4-zCWarning: The aligment of the read is longer than the actual read %sz%.5fZMRz,Read: %s already has a MR tag, can't rescalef)"logging getLogger__name__query mapdamageZalignZget_coordinatesZgetrnametidZfetchminmaxupperZalign_with_qualZcigarZqquallen is_reverseseqZrevcomprHrIr6rr rMwarningqnamerqualr Zhas_tagr/Zset_tag)bamreadrefr"r5debugr4loggerZraw_seqZ coordinateZchromZrefseqrgrjZ length_readZ length_alignZnew_qualZ pos_on_readZnumber_of_rescaled_basesrRr0r1rLZpdamZpseqZnewpZnewqrrr rescale_qual_reads`         $          rpc Cs |s$tt}|d|j|jft}t|j}|rBd}nd}tj|j||d}t |j |j |j d}t } d} d} |D]} | jrn| js|s|d| jn| jrV| r|s|d d } | js| jr| j| jkr| j| jkrt|| ||| d |d } nL| jrL| jsL| j| jkrL| j| jkrLt|| ||| d |d } n| d 7} nt|| ||| |d} || q~| dkr|s|d| | d| dks| d| dkrtdt| |||jst | |s|!dt|fdS)a~ ref a pysam fasta ref file bam_filename name of a BAM/SAM file to read fi file containing the csv with correction probabilities reflengths dictionary with the reference lengths options options from the command line parsing Iterates through BAM file, makes a new BAM file with rescaled qualities. zRescaling BAM: '%s' -> '%s'Zwhwb)template)rrTrzCCannot rescale base PHRED scores for read '%s'; no scores assigned.zUWarning! Assuming the pairs are non-overlapping, facing inwards and correctly paired.Fr-)r4rnr,)rnz9Number of non-rescaled reads due to improper pairing: %dr:r>r<r@zeQualities for T.C and A.G transitions should not change in the rescaling, please contact the authors.z!Rescaling completed in %f secondsN)"r\r]r^infofilenameZ rescale_outtimepysamZSamfiler&rrrrKZ is_unmappedrjrhriZ is_pairedrfZmate_is_reverseZpnextr2raZmrnmrpwriterrrTclosequietrVrn) rmoptionsrnroZ start_timerkZ write_modeZbam_outr"r5Z first_pairZnumber_of_non_proper_pairsZhitrrr rescale_qualsV      ,,  $ r{)r')Fr')F)rrrr`rv itertoolsrr\rur rr&r6rKrMrTrVrpr{rrrr s$ 6 ! R