fmSc@sddlZddlmZddlmZejdrMejdZndZdeZdZdZ d Z d Z d Z d Z d ZedZdS(iN(tGenome(tenvirontCISTEMATIC_ROOTs /proj/genomes"%s/C_familiaris/cfamiliaris.genedbc Csg}tdd|}t|d}|j}x!|D]}|j|jq:Wtj|d}t|} | dkrdGHnd|GHtdt|fd } | j || j |j ||d dS( Nt cfamiliaristdbFiletrtis#Problems reading sequence from fileswriting to file %ss%s%stwtfile( RtopentreadlinetappendtstriptstringtjointlentcisRoottwritetclosetaddChromosomeEntry( tdbtchromIDt chromPathtchromOuttseqArraytcfGenometinFiletlinetseqtseqLentoutFile((sh/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/cfamiliaris.pytloadChromosome+s       c Cslg}g}tdd|}t|d}|jx|D] }|jd}|djdkroq>n|djd}|d } | d ks>| |krq>n|j| t|d d } t|d d } |d } |d j} | dkrd} nd} d| f}d }|j|| | | | d|fq>Wdt|GH|j|dS(s< FIXME - NEED TO DEAL WITH ALTERNATIVE SPLICING ENTRIES RRRs i tGENEi t:iRiiit+tFtRtgenesAdding %d gene entriesN( RR R tsplitR R tintRtaddGeneEntryBatch(RtgFilet geneEntriest alreadySeenRtgeneFileRtcolstnametgidtstarttstoptsensetchromtgeneIDt gidVersion((sh/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/cfamiliaris.pytloadGeneEntries?s4        &c CsWg}tdd|}t|d}|jx|D]}|jd}|djdkriq8n|d}|d jd }|d } | d krq8nt|d d } t|dd } |d} |d j} | dkrd} nd} d| f}d }|j||| | | | |fq8Wdt|GH|j|dS(sI Load gene features such as CDS, UTR, and PSEUDO from the gene file. RRRs i tCDStUTRtPSEUDOi R!iRiiiR"R#R$sAdding %d feature entriesN(sCDSR8sPSEUDO( RR R R&R R'R RtaddFeatureEntryBatch(RR)tfeatureEntriesRt featureFileRR-tfTypeR.R/R0R1R2R3R4R5((sh/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/cfamiliaris.pytloadGeneFeaturescs2         &c Csg}t|d}tdd|}x|D]z}yj|jd}|d}|d}t|dkr|jd|ftj|jddfnWq.q.Xq.Wd t|GH|j|dS( NRRRs iit'tpsAdding %d annotations( R RR&RR R treplaceR taddAnnotationBatch( Rt annotPathtgeneAnnotationst annotFileRRR-tlocIDtgeneDesc((sh/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/cfamiliaris.pytloadGeneAnnotationss   5c Cstdd|}t|d}t|d}t}i}g}xR|D]J} | ddkrL| jd} | d| djf|| dtd?6d@tdA6dBtdC6dDtdE6dFtdG6dHtdI6dJtdK6dLtdM6dNtdO6dPtdQ6}dR|GHt|dSGHt||dTGHt||x8|jD]*}dU|GHt||||dV|q WdWGHt|dX|GHdS(YNs%s/download/seq_gene.mds%s/download/chr1.fat1s%s/download/chr2.fat2s%s/download/chr3.fat3s%s/download/chr4.fat4s%s/download/chr5.fat5s%s/download/chr6.fat6s%s/download/chr7.fat7s%s/download/chr8.fat8s%s/download/chr9.fat9s%s/download/chr10.fat10s%s/download/chr11.fat11s%s/download/chr12.fat12s%s/download/chr13.fat13s%s/download/chr14.fat14s%s/download/chr15.fat15s%s/download/chr16.fat16s%s/download/chr17.fat17s%s/download/chr18.fat18s%s/download/chr19.fat19s%s/download/chr20.fat20s%s/download/chr21.fat21s%s/download/chr22.fat22s%s/download/chr23.fat23s%s/download/chr24.fat24s%s/download/chr25.fat25s%s/download/chr26.fat26s%s/download/chr27.fat27s%s/download/chr28.fat28s%s/download/chr29.fat29s%s/download/chr30.fat30s%s/download/chr31.fat31s%s/download/chr32.fat32s%s/download/chr33.fat33s%s/download/chr34.fat34s%s/download/chr35.fat35s%s/download/chr36.fat36s%s/download/chr37.fat37s%s/download/chr38.fat38s%s/download/chrX.fatXs%s/download/chrUn.fatUnsCreating database %ssAdding gene entriessAdding gene featuressLoading chromosome %ss/C_familiaris/chromo%s.binsCreating IndicessFinished creating database %s(RRbR6R>tkeysRRd(RtgenePathtchromosR((sh/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/cfamiliaris.pyt buildDogDBsj                                             (R tcistematic.genomesRtosRtgetRtgeneDBRR6R>RHR`RbRdR(((sh/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/cfamiliaris.pyts    $ #  *