fmSc@sddlZddlmZddlmZddlmZejdr]ejdZndZdeZ dZ d Z d Z d Z d Zd Ze dZdS(iN(tGenome(t geneinfoDB(tenvirontCISTEMATIC_ROOTs /proj/genomes%s/G_gallus/ggallus.genedbc Csg}tdd|}t|d}|j}x!|D]}|j|jq:Wtj|d}t|} | dkrdGHnd|GHtdt|fd } | j || j |j ||d dS( NtggallustdbFiletrtis#Problems reading sequence from fileswriting to file %ss%s%stwtfile( RtopentreadlinetappendtstriptstringtjointlentcisRoottwritetclosetaddChromosomeEntry( tdbtchromIDt chromPathtchromOuttseqArraytggGenometinFiletlinetseqtseqLentoutFile((sd/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/ggallus.pytloadChromosome,s       c Cs~g}g}tdd|}t|d}|jx$|D]}d|krVq>n|jd}|djdkrq>n|djd }|d } | d ks>| |krq>n|j| t|d d } t|d d } |d} |d j} | dkr d} nd} d| f}d }|j|| | | | d|fq>Wdt|GH|j|dS(s< FIXME - NEED TO DEAL WITH ALTERNATIVE SPLICING ENTRIES RRRt|s i tGENEi t:iRiiit+tFtRtgenesAdding %d gene entriesN( RR R tsplitR R tintRtaddGeneEntryBatch(RtgFilet geneEntriest alreadySeenRtgeneFileRtcolstnametgidtstarttstoptsensetchromtgeneIDt gidVersion((sd/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/ggallus.pytloadGeneEntries@s8         &c Csig}tdd|}t|d}|jx|D] }d|krPq8n|jd}|djdkr{q8n|d}|d jd }|d } | d krq8nt|dd } t|dd } |d} |d j} | dkr d} nd} d| f}d }|j||| | | | |fq8Wdt|GH|j|dS(sI Load gene features such as CDS, UTR, and PSEUDO from the gene file. RRRR!s i tCDStUTRtPSEUDOi R#iRiiiR$R%R&sAdding %d feature entriesN(sCDSsUTRsPSEUDO( RR R R(R R)R RtaddFeatureEntryBatch(RR+tfeatureEntriesRt featureFileRR/tfTypeR0R1R2R3R4R5R6R7((sd/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/ggallus.pytloadGeneFeaturesgs6          &c Cstdd|}t|d}t|d}t}i}g}xR|D]J} | ddkrL| jd} | d| djf|| d  )    T cCs#tdd|}|j|dS(NRR(Rt createGeneDB(RR((sd/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/ggallus.pyt createDBFilescCs tdd|}|jdS(NRR(Rt createIndices(RR((sd/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/ggallus.pytcreateDBindicesscCs:dt}dt}dt}i9dtd6dtd6dtd 6d td 6d td 6dtd6dtd6dtd6dtd6dtd6dtd6dtd6dtd6dtd6d td!6d"td#6d$td%6d&td'6d(td)6d*td+6d,td-6d.td/6d0td16d2td36d4td56d6td76d8td96d:td;6d<td=6d>td?6d@tdA6dBtdC6dDtdE6dFtdG6dHtdI6dJtdK6dLtdM6dNtdO6dPtdQ6dRtdS6dTtdU6dVtdW6dXtdY6dZtd[6d\td]6d^td_6d`tda6dbtdc6ddtde6dftdg6dhtdi6djtdk6dltdm6dntdo6dptdq6drtds6dttdu6}dv|GHt|dwGHt||dxGHt||x8|jD]*}dy|GHt||||dz|qWd{GHt|||d|GHt|d}|GHdS(~Ns%s/download/seq_gene.mds%s/download/GO.terms_and_idss%s/download/gene2gos%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/chr32.fat32s%s/download/chrW.fatWs%s/download/chrZ.fatZs%s/download/chrM.fatMs"%s/download/chrE22C19W28_E50C23.fatE22C19W28_E50C23s%s/download/chrE64.fatE64s%s/download/chr1_random.fat1_randoms%s/download/chr2_random.fat2_randoms%s/download/chr4_random.fat4_randoms%s/download/chr5_random.fat5_randoms%s/download/chr6_random.fat6_randoms%s/download/chr7_random.fat7_randoms%s/download/chr8_random.fat8_randoms%s/download/chr10_random.fat 10_randoms%s/download/chr11_random.fat 11_randoms%s/download/chr12_random.fat 12_randoms%s/download/chr13_random.fat 13_randoms%s/download/chr16_random.fat 16_randoms%s/download/chr17_random.fat 17_randoms%s/download/chr18_random.fat 18_randoms%s/download/chr20_random.fat 20_randoms%s/download/chr22_random.fat 22_randoms%s/download/chr25_random.fat 25_randoms%s/download/chr28_random.fat 28_randoms%s/download/chrUn_random.fat Un_randoms%s/download/chrW_random.fatW_randoms%s/download/chrE64_random.fat E64_randoms%s/download/chrZ_random.fatZ_randoms)%s/download/chrE22C19W28_E50C23_random.fatE22C19W28_E50C23_randomsCreating database %ssAdding gene entriessAdding gene featuressLoading chromosome %ss/G_gallus/chromo%s.binsAdding gene ontologysCreating IndicessFinished creating database %s(RR\R8R@tkeysR RZR^(RtgenePathRKRJtchromosR((sd/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/ggallus.pytbuildChickenDBs                                                                (Rtcistematic.genomesRtcistematic.core.geneinfoRtosRtgetRtgeneDBR R8R@RZR\R^R(((sd/woldlab/castor/data00/home/georgi/code/erange-4.0a-BAM-2014-05-09-fix/cistematic/genomes/ggallus.pyts    ' & &