a d @sVddlZddlZddlZddlZddlZddZddZddZd d d Zd d Z dS)NcCs\|jr dnd}t||d||tt| dt||d||tdt|ddS)z6 record basae composition in external genomic regions -+5pr3pN) is_reverse _update_tablerangelen)readchrombeforeZaftercompstdr4lib/python3.9/site-packages/mapdamage/composition.pycount_ref_comps$rcCstd|j}}|jr$dtj|}}t||d||td|dt||d|t|td| dddS)z= record base composition of read, discard marked nucleotides rrrrrN)queryr mapdamageseqZrevcomprr reversed)r r lengthrrrrrrcount_read_comps  "rcCs4t||D]$\}}||vr |||d7<q dS)Nr)zip)tableZsequenceindicesindexntrrrrsrFc sdddddgd}ztjdd|gtjd}|d}|D]l}|}dt|dd<dt|d d<d t|d d <d t|d d <qBWnDttfy}z(t j d|ft dWYd}~n d}~00t }|D]} t| t|| <q|dkr6St|d} fdd|D} | d|d| d| d| dS)zv Gets the basecomposition of all the sequences in filename and returns the value to destination if given. r)ACGTZseqtkr)stdoutrr r!r"zError: Seqtk failed: %s rNFwcsg|]}t|qSr)str).0ibasesrr ;z!get_base_comp.., ) subprocessPopenPIPEZ communicate splitlinessplitintOSError ValueErrorsysstderrwriteexitsumvaluesfloatopenjoinclose) filenameZ destinationZalpprocoutliZtabserrorZba_subafoZvalsrr,r get_base_comps0      rKc Cs\t|}d}|D]&}|}|}|r2|}d}q|}qi}t||D]\}} | ||<qF|S)zF Read the base compition from a file created by get_base_comp TF)rArstripr6r) rDZfhZ first_linerGZlpheaderbodyr-rIZperrrrread_base_comp@s rO)F) r itertoolsZcsvr2r:rrrrKrOrrrrs  !