VNc @s;ddlZddlZddlZddlmZddlmZddlmZddlZddl m Z ddl m Z ddl Z dZdZd efd YZd Zd Zd ZedZedZeedZgdZddZddZddZddZdZddeeedddee edd Z!ddeeedddee edd Z"dZ#dZ$deedd d!Z%ed d"Z&d#Z'd$Z(ieged%d&Z)dde ieedee d' Z*d(Z+gd)e ddd*Z,dS(+iN(tstrftime(tarray(t defaultdict(t geneinfoDB(tGenomegffffff@g@t ErangeErrorcBseZRS((t__name__t __module__(((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pyRscCs1idd6dd6dd6dd6dd6}||S(NtTtAtCtGtN((tbasetrevComp((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pytgetReverseComplements   cCs7tt}x|D]}||cd7|}nX|S(N(t getbooleanR9RBRCt ValueError(RDRERFRGRH((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pytgetConfigBoolOptionws  cCs4y|j|}Wntjk r/g}nX|S(N(RR9RB(RDRERH((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pytgetAllConfigSectionOptionss  iiiic CsVt|} | j}t|||||||||| | | | }| j|S(s returns a dictionary containing a list of merged overlapping regions by chromosome; can optionally filter regions that have a scoreField fewer than minHits. Can also optionally return the label of each region, as well as the peak, if supplied (peakPos and peakHeight should be the last 2 fields). Can return the top regions based on score if higher than minHits. (Rt readlinestgetMergedRegionsFromListR(tregionfilenametmaxDisttminHitstverboset keepLabelt fullChromt chromFieldt scoreFieldtpadtcompacttdoMergetkeepPeakt returnToptinfiletlinestregions((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pytgetMergedRegionss   c (Csi} d}d}d| ko/t|knrg}x|D]~}|ddkrd|krld}nd|krAd}qAqAn|jjd}t||j}|j|qAW|jd| } || }||kr|}qnd}t|}d}x|D] }|ddkr^d|krCd}nd|krd}qqn|jjd}|dkryt||j}Wnttfk rqnX||krqqn| r||jd\}}|jd \}}t|}t|}n|dkryt j || d}||}t||d|}t||d |}n<|d}|d}t|d |}t|d |}|s|d }n| rt|d ||}t|d||}n|| krg| |'ssmerged %d timessreturning %d regions(tlentstriptsplittfloatR8tsortRt IndexErrorROtstringtjointFalsetrangeRktstoptregionsOverlaptregionsAreWithinDistancetpeakPost peakHeighttabstlengthtlabeltTruetRegion((t regionListRURVRWRXRYRZR[R\R]R^R_R`Rct hasPvaluethasShifttscorest regionEntrytfieldsthitstminScoret mergeCounttcounttchromtpostfronttbackRkRyRR|R}tmergedtindextregiontrstarttrstoptrpeakPost rpeakHeightt regionCount((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pyRSs "                             -                  cCs||kr||}}n||kr8||}}n||koO|knp||kok|knp||ko|knp||ko|kSS(N((RkRyRR((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pyRz0s   cCsb||kr||}}n||kr8||}}nt|||kpat|||kS(N(R~(RkRyRRRU((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pyR{:s   iiKcCs|dkr-t|||d|d|}nt|||||||\}} } } tddg|} xitd|dD]T} || dd|| dd|| d|| d|| dd | | 0 bp, e.g. X, return only the starting X bp from CDS. If lengthCDS < 0bp, return only the last X bp from CDS. isAgetLocusByChromDict: asked for no sequence - returning empty dictsRgetLocusByChromDict: asked for only upstream and downstream - returning empty dictsPgetLocusByChromDict: asked for partial CDS but not useCDS - returning empty dictsRgetLocusByChromDict: asked for TSS spanning and partial CDS - returning empty dictsigetLocusByChromDict: asked for discontinuous partial CDS from start and downstream - returning empty dictsfgetLocusByChromDict: asked for discontinuous partial CDS from stop and upstream - returning empty dictRiiRi(R&tgetGeneFeaturesR8RsR~tleftGeneDistancetmaxtrightGeneDistance(R&tupstreamt downstreamtuseCDSRRtupstreamSpanTSSt lengthCDSt keepSensetadjustToNeighbortlocusByChromDictR1RRRRRRRkRyRtgstarttgstoptglentdistancetnextGene((sE/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM/commoncode.pytgetLocusByChromDict3s                      #! c Cs|j}t|dkrg}x|D]}x||D]}|j}||krg|j|n||krg|||D]6}x-||D]!}|| }dg|| |s\           *% ( i