a ŽÞd ã@sLddlZddlZdd„Zdd„Zdd„Zdd „Zd d „Zd d „Zdd„ZdS)éNcCs,|jr |jn|j}|jr|jn|j}||fS)z3 return external coordinates of aligned read bases )Ú is_reverseZaendÚpos)ÚreadZfivepZthreep©rú.lib/python3.9/site-packages/mapdamage/align.pyÚget_coordinatessrc Cs\t|ƒ}t|ƒ}td||ƒ}t||||ƒ}| |||¡ ¡} | |||¡ ¡} | | fS)zk return reference sequences before and after the read check for extremities and return what is available r)ÚminÚmaxZfetchÚupper) ZcoordZchromZ reflengthsÚlengthÚrefZ coord_minZ coord_maxZ pos_beforeZ pos_afterÚbeforeZafterrrrÚ get_aroundsrcCspt|ƒ}t|dƒD]\}}dg||||…<qt|ƒ}t|dƒD]\}}dg||||…<q@d |¡d |¡fS)zš insert gaps according to the cigar string deletion: gaps to be inserted into read sequences, insertions: gaps to be inserted into reference sequence éú-éÚ)ÚlistÚ parse_cigarÚjoin)Ú cigarlistÚseqr ÚlrefÚnbrÚidxÚlreadrrrÚalign)src CsÐt|ƒ}t|dƒD]\}}dg||||…<qt|ƒ}t|ƒ} t|dƒD],\}}dg||||…<dg|| ||…<qHt| ƒD]4\}} t| ƒd|kr~||dkr~d||<d||<q~d |¡d | ¡d |¡fS)z¨ insert gaps according to the cigar string deletion: gaps to be inserted into read sequences and qualities, insertions: gaps to be inserted into reference sequence rrré!ÚNr)rrÚ enumerateÚordr) rrZqualZ thresholdr rrrrZlqualZscorerrrÚalign_with_qual8s r!c Cs’|jr dnd}||||}tt|ƒ||ƒD]^\} } } | dvr.| dvr.| dkrd|| | d7<| | kr.d| | f} || | d7<q.dS)z~ count mismatches using aligned reference and read, must be redone since N in reference were randomly replaced by any bases rú+zACGT-rz%s>%sN)rÚzipÚrange) rrZrefseqr r ÚtabÚendÚstdZsubtableÚiZnt_seqZnt_refZmutrrrÚget_misOs r)cCsDd}g}d}|D].\}}||kr.| ||g¡||vr||7}q|S)z| for a specific operation (mismach, match, insertion, deletion... see above) return occurences and index in the alignment r)rrréé)Úappend)rZopeZtlengthZ coordinateZoplistZ operationr rrrras  rcsl‡‡fdd„}|jrdnd}tj |jd¡D]:\}}|dkrL|jrFdnd}n|jrVdnd}||||ƒq,d S) z* record soft clipped bases at extremities cs6tdt|ˆƒƒD] }ˆ||d|d7<qdS)NrÚSr)r$r)r&r'Úbasesr(©r r%rrÚ update_tablersz*record_soft_clipping..update_tablerr"érZ3pZ5pN)rÚ mapdamagerrZcigar)rr%r r0ZstrandZnbasesrr&rr/rÚrecord_soft_clippingpsr3) Ú itertoolsr2rrrr!r)rr3rrrrÚs