bio-ucsc-api 0.3.1 → 0.4.0
Sign up to get free protection for your applications and to get access to all the features.
- data/Gemfile +4 -3
- data/Gemfile.lock +31 -20
- data/README.md +224 -0
- data/Rakefile +5 -8
- data/VERSION +1 -1
- data/bio-ucsc-api.gemspec +39 -122
- data/lib/bio-ucsc.rb +10 -5
- data/lib/bio-ucsc/ailmel1.rb +5 -3
- data/lib/bio-ucsc/anocar2.rb +7 -6
- data/lib/bio-ucsc/anogam1.rb +7 -5
- data/lib/bio-ucsc/anogam1/chaindm3.rb +11 -8
- data/lib/bio-ucsc/anogam1/chaindm3link.rb +11 -8
- data/lib/bio-ucsc/anogam1/est.rb +11 -8
- data/lib/bio-ucsc/anogam1/gap.rb +11 -8
- data/lib/bio-ucsc/anogam1/gold.rb +11 -8
- data/lib/bio-ucsc/anogam1/intronest.rb +11 -8
- data/lib/bio-ucsc/anogam1/mrna.rb +11 -8
- data/lib/bio-ucsc/anogam1/rmsk.rb +10 -7
- data/lib/bio-ucsc/apimel2.rb +5 -3
- data/lib/bio-ucsc/apimel2/chaindm2.rb +11 -8
- data/lib/bio-ucsc/apimel2/chaindm2link.rb +11 -8
- data/lib/bio-ucsc/aplcal1.rb +5 -3
- data/lib/bio-ucsc/aplcal1/chaincaepb1.rb +11 -8
- data/lib/bio-ucsc/aplcal1/chaincaepb1link.rb +11 -8
- data/lib/bio-ucsc/aplcal1/chaincaerem2.rb +11 -8
- data/lib/bio-ucsc/aplcal1/chaincaerem2link.rb +11 -8
- data/lib/bio-ucsc/aplcal1/chaincb3.rb +11 -8
- data/lib/bio-ucsc/aplcal1/chaincb3link.rb +11 -8
- data/lib/bio-ucsc/aplcal1/chaince6.rb +11 -8
- data/lib/bio-ucsc/aplcal1/chaince6link.rb +11 -8
- data/lib/bio-ucsc/aplcal1/gap.rb +11 -8
- data/lib/bio-ucsc/aplcal1/gold.rb +11 -8
- data/lib/bio-ucsc/aplcal1/rmsk.rb +10 -7
- data/lib/bio-ucsc/bostau4.rb +5 -3
- data/lib/bio-ucsc/braflo1.rb +5 -3
- data/lib/bio-ucsc/braflo1/chaingalgal3.rb +11 -8
- data/lib/bio-ucsc/braflo1/chaingalgal3link.rb +11 -8
- data/lib/bio-ucsc/braflo1/chainhg18.rb +11 -8
- data/lib/bio-ucsc/braflo1/chainhg18link.rb +11 -8
- data/lib/bio-ucsc/braflo1/chainmm9.rb +11 -8
- data/lib/bio-ucsc/braflo1/chainmm9link.rb +11 -8
- data/lib/bio-ucsc/braflo1/chainpetmar1.rb +11 -8
- data/lib/bio-ucsc/braflo1/chainpetmar1link.rb +11 -8
- data/lib/bio-ucsc/braflo1/gap.rb +11 -8
- data/lib/bio-ucsc/braflo1/gold.rb +11 -8
- data/lib/bio-ucsc/caejap1.rb +5 -3
- data/lib/bio-ucsc/caejap1/chaince6.rb +11 -8
- data/lib/bio-ucsc/caejap1/chaince6link.rb +11 -8
- data/lib/bio-ucsc/caejap1/gap.rb +11 -8
- data/lib/bio-ucsc/caejap1/gold.rb +11 -8
- data/lib/bio-ucsc/caepb2.rb +6 -4
- data/lib/bio-ucsc/caepb2/chaince6.rb +11 -8
- data/lib/bio-ucsc/caepb2/chaince6link.rb +11 -8
- data/lib/bio-ucsc/caepb2/gap.rb +11 -8
- data/lib/bio-ucsc/caepb2/gold.rb +11 -8
- data/lib/bio-ucsc/caerem3.rb +5 -3
- data/lib/bio-ucsc/caerem3/chaince6.rb +11 -8
- data/lib/bio-ucsc/caerem3/chaince6link.rb +11 -8
- data/lib/bio-ucsc/caerem3/gap.rb +11 -8
- data/lib/bio-ucsc/caerem3/gold.rb +11 -8
- data/lib/bio-ucsc/caljac3.rb +7 -4
- data/lib/bio-ucsc/canfam2.rb +5 -3
- data/lib/bio-ucsc/canfam2/chainbostau4.rb +11 -8
- data/lib/bio-ucsc/canfam2/chainbostau4link.rb +11 -8
- data/lib/bio-ucsc/canfam2/chainmm9.rb +11 -8
- data/lib/bio-ucsc/canfam2/chainmm9link.rb +11 -8
- data/lib/bio-ucsc/canfam2/chainrn4.rb +11 -8
- data/lib/bio-ucsc/canfam2/chainrn4link.rb +11 -8
- data/lib/bio-ucsc/canfam2/chainself.rb +11 -8
- data/lib/bio-ucsc/canfam2/chainselflink.rb +11 -8
- data/lib/bio-ucsc/canfam2/est.rb +11 -8
- data/lib/bio-ucsc/canfam2/gap.rb +11 -8
- data/lib/bio-ucsc/canfam2/gold.rb +11 -8
- data/lib/bio-ucsc/canfam2/intronest.rb +11 -8
- data/lib/bio-ucsc/canfam2/mrna.rb +11 -8
- data/lib/bio-ucsc/canfam2/rmsk.rb +10 -7
- data/lib/bio-ucsc/cavpor3.rb +6 -4
- data/lib/bio-ucsc/cb3.rb +5 -3
- data/lib/bio-ucsc/cb3/chaincaepb1.rb +11 -8
- data/lib/bio-ucsc/cb3/chaincaepb1link.rb +11 -8
- data/lib/bio-ucsc/cb3/chaincaerem2.rb +11 -8
- data/lib/bio-ucsc/cb3/chaincaerem2link.rb +11 -8
- data/lib/bio-ucsc/cb3/chaince6.rb +11 -8
- data/lib/bio-ucsc/cb3/chaince6link.rb +11 -8
- data/lib/bio-ucsc/cb3/chainpripac1.rb +11 -8
- data/lib/bio-ucsc/cb3/chainpripac1link.rb +11 -8
- data/lib/bio-ucsc/cb3/est.rb +11 -8
- data/lib/bio-ucsc/cb3/gap.rb +11 -8
- data/lib/bio-ucsc/cb3/gold.rb +11 -8
- data/lib/bio-ucsc/cb3/intronest.rb +11 -8
- data/lib/bio-ucsc/cb3/mrna.rb +11 -8
- data/lib/bio-ucsc/cb3/rmsk.rb +10 -7
- data/lib/bio-ucsc/ce6.rb +8 -4
- data/lib/bio-ucsc/ce6/chaincaejap1.rb +11 -8
- data/lib/bio-ucsc/ce6/chaincaejap1link.rb +11 -8
- data/lib/bio-ucsc/ce6/chaincaepb2.rb +11 -8
- data/lib/bio-ucsc/ce6/chaincaepb2link.rb +11 -8
- data/lib/bio-ucsc/ce6/chaincaerem3.rb +11 -8
- data/lib/bio-ucsc/ce6/chaincaerem3link.rb +11 -8
- data/lib/bio-ucsc/ce6/chaincb3.rb +11 -8
- data/lib/bio-ucsc/ce6/chaincb3link.rb +11 -8
- data/lib/bio-ucsc/ce6/chainpripac1.rb +11 -8
- data/lib/bio-ucsc/ce6/chainpripac1link.rb +11 -8
- data/lib/bio-ucsc/ce6/chainself.rb +11 -8
- data/lib/bio-ucsc/ce6/chainselflink.rb +11 -8
- data/lib/bio-ucsc/ce6/est.rb +11 -8
- data/lib/bio-ucsc/ce6/gap.rb +11 -8
- data/lib/bio-ucsc/ce6/gold.rb +11 -8
- data/lib/bio-ucsc/ce6/intronest.rb +11 -8
- data/lib/bio-ucsc/ce6/mrna.rb +11 -8
- data/lib/bio-ucsc/ce6/rmsk.rb +11 -8
- data/lib/bio-ucsc/ce6/t25mersrepeats.rb +7 -4
- data/lib/bio-ucsc/ci2.rb +6 -3
- data/lib/bio-ucsc/danrer7.rb +6 -3
- data/lib/bio-ucsc/{apimel2/db_connection.rb → db_connector.rb} +33 -23
- data/lib/bio-ucsc/dm3.rb +6 -4
- data/lib/bio-ucsc/dm3/chainanogam1.rb +11 -8
- data/lib/bio-ucsc/dm3/chainanogam1link.rb +11 -8
- data/lib/bio-ucsc/dm3/chainapimel3.rb +11 -8
- data/lib/bio-ucsc/dm3/chainapimel3link.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindp3.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindp3link.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindroana2.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindroana2link.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindroere1.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindroere1link.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindroper1.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindroper1link.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindrosec1.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindrosec1link.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindrosim1.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindrosim1link.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindroyak2.rb +11 -8
- data/lib/bio-ucsc/dm3/chaindroyak2link.rb +11 -8
- data/lib/bio-ucsc/dm3/est.rb +11 -8
- data/lib/bio-ucsc/dm3/gap.rb +11 -8
- data/lib/bio-ucsc/dm3/gold.rb +11 -8
- data/lib/bio-ucsc/dm3/intronest.rb +11 -8
- data/lib/bio-ucsc/dm3/mrna.rb +11 -8
- data/lib/bio-ucsc/dm3/rmsk.rb +11 -8
- data/lib/bio-ucsc/dp3.rb +6 -4
- data/lib/bio-ucsc/dp3/chaindm3.rb +11 -8
- data/lib/bio-ucsc/dp3/chaindm3link.rb +11 -8
- data/lib/bio-ucsc/dp3/est.rb +11 -8
- data/lib/bio-ucsc/dp3/gap.rb +11 -8
- data/lib/bio-ucsc/dp3/gold.rb +11 -8
- data/lib/bio-ucsc/dp3/intronest.rb +11 -8
- data/lib/bio-ucsc/dp3/mrna.rb +11 -8
- data/lib/bio-ucsc/dp3/rmsk.rb +10 -7
- data/lib/bio-ucsc/droana2.rb +6 -4
- data/lib/bio-ucsc/droere1.rb +6 -4
- data/lib/bio-ucsc/drogri1.rb +6 -4
- data/lib/bio-ucsc/dromoj2.rb +7 -5
- data/lib/bio-ucsc/droper1.rb +7 -5
- data/lib/bio-ucsc/drosec1.rb +6 -4
- data/lib/bio-ucsc/drosim1.rb +6 -4
- data/lib/bio-ucsc/drosim1/chaindm3.rb +11 -8
- data/lib/bio-ucsc/drosim1/chaindm3link.rb +11 -8
- data/lib/bio-ucsc/drosim1/est.rb +11 -8
- data/lib/bio-ucsc/drosim1/gap.rb +11 -8
- data/lib/bio-ucsc/drosim1/gold.rb +11 -8
- data/lib/bio-ucsc/drosim1/intronest.rb +11 -8
- data/lib/bio-ucsc/drosim1/mrna.rb +11 -8
- data/lib/bio-ucsc/drosim1/rmsk.rb +10 -7
- data/lib/bio-ucsc/drovir2.rb +6 -4
- data/lib/bio-ucsc/droyak2.rb +6 -4
- data/lib/bio-ucsc/droyak2/chaindm3.rb +11 -8
- data/lib/bio-ucsc/droyak2/chaindm3link.rb +11 -8
- data/lib/bio-ucsc/droyak2/est.rb +11 -8
- data/lib/bio-ucsc/droyak2/gap.rb +11 -8
- data/lib/bio-ucsc/droyak2/gold.rb +11 -8
- data/lib/bio-ucsc/droyak2/intronest.rb +11 -8
- data/lib/bio-ucsc/droyak2/mrna.rb +11 -8
- data/lib/bio-ucsc/droyak2/rmsk.rb +10 -7
- data/lib/bio-ucsc/equcab2.rb +7 -4
- data/lib/bio-ucsc/equcab2/est.rb +11 -8
- data/lib/bio-ucsc/equcab2/gap.rb +11 -8
- data/lib/bio-ucsc/equcab2/gold.rb +11 -8
- data/lib/bio-ucsc/equcab2/intronest.rb +11 -8
- data/lib/bio-ucsc/equcab2/mrna.rb +11 -8
- data/lib/bio-ucsc/equcab2/rmsk.rb +10 -7
- data/lib/bio-ucsc/felcat4.rb +7 -4
- data/lib/bio-ucsc/file.rb +10 -0
- data/lib/bio-ucsc/file/bytequeue.rb +28 -0
- data/lib/bio-ucsc/file/twobit.rb +174 -0
- data/lib/bio-ucsc/fr2.rb +5 -3
- data/lib/bio-ucsc/fr2/chaingalgal3.rb +11 -8
- data/lib/bio-ucsc/fr2/chaingalgal3link.rb +11 -8
- data/lib/bio-ucsc/fr2/chaingasacu1.rb +11 -8
- data/lib/bio-ucsc/fr2/chaingasacu1link.rb +11 -8
- data/lib/bio-ucsc/fr2/chainmm9.rb +11 -8
- data/lib/bio-ucsc/fr2/chainmm9link.rb +11 -8
- data/lib/bio-ucsc/fr2/chainorylat2.rb +11 -8
- data/lib/bio-ucsc/fr2/chainorylat2link.rb +11 -8
- data/lib/bio-ucsc/fr2/est.rb +11 -8
- data/lib/bio-ucsc/fr2/gap.rb +11 -8
- data/lib/bio-ucsc/fr2/gold.rb +11 -8
- data/lib/bio-ucsc/fr2/intronest.rb +11 -8
- data/lib/bio-ucsc/fr2/mrna.rb +11 -8
- data/lib/bio-ucsc/fr2/rmsk.rb +10 -7
- data/lib/bio-ucsc/galgal3.rb +8 -5
- data/lib/bio-ucsc/galgal3/chainanocar1.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainanocar1link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainbraflo1.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainbraflo1link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chaincavpor3.rb +11 -8
- data/lib/bio-ucsc/galgal3/chaincavpor3link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chaindanrer4.rb +11 -8
- data/lib/bio-ucsc/galgal3/chaindanrer4link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainfr2.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainfr2link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chaingasacu1.rb +11 -8
- data/lib/bio-ucsc/galgal3/chaingasacu1link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainmm9.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainmm9link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainornana1.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainornana1link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainpetmar1.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainpetmar1link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainrn4.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainrn4link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chaintaegut1.rb +11 -8
- data/lib/bio-ucsc/galgal3/chaintaegut1link.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainxentro2.rb +11 -8
- data/lib/bio-ucsc/galgal3/chainxentro2link.rb +11 -8
- data/lib/bio-ucsc/galgal3/est.rb +11 -8
- data/lib/bio-ucsc/galgal3/gap.rb +11 -8
- data/lib/bio-ucsc/galgal3/gold.rb +11 -8
- data/lib/bio-ucsc/galgal3/intronest.rb +11 -8
- data/lib/bio-ucsc/galgal3/mrna.rb +11 -8
- data/lib/bio-ucsc/galgal3/rmsk.rb +10 -7
- data/lib/bio-ucsc/gasacu1.rb +6 -4
- data/lib/bio-ucsc/gasacu1/chainanocar1.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chainanocar1link.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chainfr2.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chainfr2link.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chaingalgal3.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chaingalgal3link.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chainmm9.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chainmm9link.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chainorylat2.rb +11 -8
- data/lib/bio-ucsc/gasacu1/chainorylat2link.rb +11 -8
- data/lib/bio-ucsc/gasacu1/est.rb +11 -8
- data/lib/bio-ucsc/gasacu1/intronest.rb +11 -8
- data/lib/bio-ucsc/gasacu1/mrna.rb +11 -8
- data/lib/bio-ucsc/gasacu1/rmsk.rb +10 -7
- data/lib/bio-ucsc/gi.rb +27 -0
- data/lib/bio-ucsc/go.rb +7 -4
- data/lib/bio-ucsc/hg18.rb +7 -4
- data/lib/bio-ucsc/hg18/chainanocar1.rb +11 -8
- data/lib/bio-ucsc/hg18/chainbostau4.rb +11 -8
- data/lib/bio-ucsc/hg18/chainbraflo1.rb +11 -8
- data/lib/bio-ucsc/hg18/chaincaljac1.rb +11 -8
- data/lib/bio-ucsc/hg18/chaincanfam2.rb +11 -8
- data/lib/bio-ucsc/hg18/chaincavpor3.rb +11 -8
- data/lib/bio-ucsc/hg18/chaindanrer5.rb +11 -8
- data/lib/bio-ucsc/hg18/chainequcab1.rb +11 -8
- data/lib/bio-ucsc/hg18/chainfelcat3.rb +11 -8
- data/lib/bio-ucsc/hg18/chainfr2.rb +12 -9
- data/lib/bio-ucsc/hg18/chaingalgal3.rb +11 -8
- data/lib/bio-ucsc/hg18/chaingasacu1.rb +11 -8
- data/lib/bio-ucsc/hg18/chainmm9.rb +11 -8
- data/lib/bio-ucsc/hg18/chainmondom4.rb +11 -8
- data/lib/bio-ucsc/hg18/chainornana1.rb +11 -8
- data/lib/bio-ucsc/hg18/chainorylat2.rb +11 -8
- data/lib/bio-ucsc/hg18/chainpantro2.rb +11 -8
- data/lib/bio-ucsc/hg18/chainpetmar1.rb +11 -8
- data/lib/bio-ucsc/hg18/chainponabe2.rb +11 -8
- data/lib/bio-ucsc/hg18/chainrhemac2.rb +11 -8
- data/lib/bio-ucsc/hg18/chainrn4.rb +11 -8
- data/lib/bio-ucsc/hg18/chainself.rb +11 -8
- data/lib/bio-ucsc/hg18/chainstrpur2.rb +11 -8
- data/lib/bio-ucsc/hg18/chaintaegut1.rb +11 -8
- data/lib/bio-ucsc/hg18/chaintetnig2.rb +11 -8
- data/lib/bio-ucsc/hg18/chainxentro2.rb +11 -8
- data/lib/bio-ucsc/hg18/gap.rb +11 -8
- data/lib/bio-ucsc/hg18/gold.rb +11 -8
- data/lib/bio-ucsc/hg18/intronest.rb +11 -8
- data/lib/bio-ucsc/hg18/rmsk.rb +10 -7
- data/lib/bio-ucsc/hg18/rmskrm327.rb +10 -7
- data/lib/bio-ucsc/hg19.rb +7 -6
- data/lib/bio-ucsc/hg19/description.rb +1 -1
- data/lib/bio-ucsc/hg19/gbcdnainfo.rb +2 -2
- data/lib/bio-ucsc/hgfixed.rb +7 -4
- data/lib/bio-ucsc/loxafr3.rb +7 -4
- data/lib/bio-ucsc/mm9.rb +7 -12
- data/lib/bio-ucsc/mm9/chainbraflo1.rb +11 -8
- data/lib/bio-ucsc/mm9/chainbraflo1link.rb +11 -8
- data/lib/bio-ucsc/mm9/chaincanfam2.rb +11 -8
- data/lib/bio-ucsc/mm9/chaincanfam2link.rb +11 -8
- data/lib/bio-ucsc/mm9/chaincavpor3.rb +11 -8
- data/lib/bio-ucsc/mm9/chaincavpor3link.rb +11 -8
- data/lib/bio-ucsc/mm9/chainfr2.rb +11 -8
- data/lib/bio-ucsc/mm9/chainfr2link.rb +11 -8
- data/lib/bio-ucsc/mm9/chaingalgal3.rb +11 -8
- data/lib/bio-ucsc/mm9/chaingalgal3link.rb +11 -8
- data/lib/bio-ucsc/mm9/chaingasacu1.rb +11 -8
- data/lib/bio-ucsc/mm9/chaingasacu1link.rb +11 -8
- data/lib/bio-ucsc/mm9/chainornana1.rb +11 -8
- data/lib/bio-ucsc/mm9/chainornana1link.rb +11 -8
- data/lib/bio-ucsc/mm9/chainorylat2.rb +11 -8
- data/lib/bio-ucsc/mm9/chainorylat2link.rb +11 -8
- data/lib/bio-ucsc/mm9/chainpetmar1.rb +11 -8
- data/lib/bio-ucsc/mm9/chainpetmar1link.rb +11 -8
- data/lib/bio-ucsc/mm9/chainponabe2.rb +11 -8
- data/lib/bio-ucsc/mm9/chainponabe2link.rb +11 -8
- data/lib/bio-ucsc/mm9/chainrhemac2.rb +11 -8
- data/lib/bio-ucsc/mm9/chainrhemac2link.rb +11 -8
- data/lib/bio-ucsc/mm9/chainrn4.rb +11 -8
- data/lib/bio-ucsc/mm9/chainrn4link.rb +11 -8
- data/lib/bio-ucsc/mm9/est.rb +11 -8
- data/lib/bio-ucsc/mm9/gap.rb +11 -8
- data/lib/bio-ucsc/mm9/gold.rb +11 -8
- data/lib/bio-ucsc/mm9/intronest.rb +11 -8
- data/lib/bio-ucsc/mm9/mrna.rb +11 -8
- data/lib/bio-ucsc/mm9/rmsk.rb +10 -7
- data/lib/bio-ucsc/mondom5.rb +7 -4
- data/lib/bio-ucsc/ornana1.rb +6 -3
- data/lib/bio-ucsc/orycun2.rb +5 -3
- data/lib/bio-ucsc/orylat2.rb +7 -5
- data/lib/bio-ucsc/oviari1.rb +6 -4
- data/lib/bio-ucsc/pantro3.rb +7 -5
- data/lib/bio-ucsc/petmar1.rb +6 -5
- data/lib/bio-ucsc/ponabe2.rb +6 -4
- data/lib/bio-ucsc/ponabe2/chainmm9.rb +11 -8
- data/lib/bio-ucsc/ponabe2/chainmm9link.rb +11 -8
- data/lib/bio-ucsc/ponabe2/chainornana1.rb +11 -8
- data/lib/bio-ucsc/ponabe2/chainornana1link.rb +11 -8
- data/lib/bio-ucsc/ponabe2/chainrhemac2.rb +11 -8
- data/lib/bio-ucsc/ponabe2/chainrhemac2link.rb +11 -8
- data/lib/bio-ucsc/ponabe2/est.rb +11 -8
- data/lib/bio-ucsc/ponabe2/gap.rb +11 -8
- data/lib/bio-ucsc/ponabe2/gold.rb +11 -8
- data/lib/bio-ucsc/ponabe2/intronest.rb +11 -8
- data/lib/bio-ucsc/ponabe2/mrna.rb +11 -8
- data/lib/bio-ucsc/ponabe2/rmsk.rb +10 -7
- data/lib/bio-ucsc/pripac1.rb +7 -5
- data/lib/bio-ucsc/pripac1/chaincaepb1.rb +11 -8
- data/lib/bio-ucsc/pripac1/chaincaepb1link.rb +11 -8
- data/lib/bio-ucsc/pripac1/chaincaerem2.rb +11 -8
- data/lib/bio-ucsc/pripac1/chaincaerem2link.rb +11 -8
- data/lib/bio-ucsc/pripac1/chaincb3.rb +11 -8
- data/lib/bio-ucsc/pripac1/chaincb3link.rb +11 -8
- data/lib/bio-ucsc/pripac1/chaince6.rb +11 -8
- data/lib/bio-ucsc/pripac1/chaince6link.rb +11 -8
- data/lib/bio-ucsc/pripac1/gap.rb +11 -8
- data/lib/bio-ucsc/pripac1/gold.rb +11 -8
- data/lib/bio-ucsc/pripac1/rmsk.rb +10 -7
- data/lib/bio-ucsc/proteome.rb +7 -4
- data/lib/bio-ucsc/reference.rb +5 -160
- data/lib/bio-ucsc/rhemac2.rb +5 -5
- data/lib/bio-ucsc/rhemac2/chainmm9.rb +11 -8
- data/lib/bio-ucsc/rhemac2/chainmm9link.rb +11 -8
- data/lib/bio-ucsc/rhemac2/chainponabe2.rb +11 -8
- data/lib/bio-ucsc/rhemac2/chainponabe2link.rb +11 -8
- data/lib/bio-ucsc/rhemac2/chainrn4.rb +11 -8
- data/lib/bio-ucsc/rhemac2/chainrn4link.rb +11 -8
- data/lib/bio-ucsc/rn4.rb +6 -8
- data/lib/bio-ucsc/rn4/chaincanfam2.rb +11 -8
- data/lib/bio-ucsc/rn4/chaincanfam2link.rb +11 -8
- data/lib/bio-ucsc/rn4/chaincavpor3.rb +11 -8
- data/lib/bio-ucsc/rn4/chaincavpor3link.rb +11 -8
- data/lib/bio-ucsc/rn4/chaindanrer4.rb +11 -8
- data/lib/bio-ucsc/rn4/chaindanrer4link.rb +11 -8
- data/lib/bio-ucsc/rn4/chainequcab1.rb +11 -8
- data/lib/bio-ucsc/rn4/chainequcab1link.rb +11 -8
- data/lib/bio-ucsc/rn4/chaingalgal3.rb +11 -8
- data/lib/bio-ucsc/rn4/chaingalgal3link.rb +11 -8
- data/lib/bio-ucsc/rn4/chainmm9.rb +11 -8
- data/lib/bio-ucsc/rn4/chainmm9link.rb +11 -8
- data/lib/bio-ucsc/rn4/chainmondom4.rb +11 -8
- data/lib/bio-ucsc/rn4/chainmondom4link.rb +11 -8
- data/lib/bio-ucsc/rn4/chainpantro2.rb +11 -8
- data/lib/bio-ucsc/rn4/chainpantro2link.rb +11 -8
- data/lib/bio-ucsc/rn4/chainrhemac2.rb +11 -8
- data/lib/bio-ucsc/rn4/chainrhemac2link.rb +11 -8
- data/lib/bio-ucsc/rn4/est.rb +11 -8
- data/lib/bio-ucsc/rn4/gap.rb +11 -8
- data/lib/bio-ucsc/rn4/gold.rb +11 -8
- data/lib/bio-ucsc/rn4/intronest.rb +11 -8
- data/lib/bio-ucsc/rn4/mrna.rb +11 -8
- data/lib/bio-ucsc/rn4/rmsk.rb +10 -7
- data/lib/bio-ucsc/saccer2.rb +6 -4
- data/lib/bio-ucsc/saccer2/est.rb +11 -8
- data/lib/bio-ucsc/saccer2/gap.rb +11 -8
- data/lib/bio-ucsc/saccer2/gold.rb +11 -8
- data/lib/bio-ucsc/saccer2/intronest.rb +11 -8
- data/lib/bio-ucsc/saccer2/mrna.rb +11 -8
- data/lib/bio-ucsc/saccer2/t2micron_est.rb +4 -4
- data/lib/bio-ucsc/saccer2/t2micron_gap.rb +4 -4
- data/lib/bio-ucsc/saccer2/t2micron_gold.rb +4 -4
- data/lib/bio-ucsc/saccer2/t2micron_intronEst.rb +4 -4
- data/lib/bio-ucsc/saccer2/t2micron_mrna.rb +4 -4
- data/lib/bio-ucsc/strpur2.rb +7 -15
- data/lib/bio-ucsc/susscr2.rb +6 -4
- data/lib/bio-ucsc/susscr2/est.rb +11 -8
- data/lib/bio-ucsc/susscr2/intronest.rb +11 -8
- data/lib/bio-ucsc/susscr2/mrna.rb +11 -8
- data/lib/bio-ucsc/table_class_detector.rb +250 -475
- data/lib/bio-ucsc/taegut1.rb +6 -4
- data/lib/bio-ucsc/taegut1/chaingalgal3.rb +14 -13
- data/lib/bio-ucsc/taegut1/chaingalgal3link.rb +13 -10
- data/lib/bio-ucsc/taegut1/est.rb +13 -10
- data/lib/bio-ucsc/taegut1/gap.rb +13 -10
- data/lib/bio-ucsc/taegut1/gold.rb +13 -10
- data/lib/bio-ucsc/taegut1/intronest.rb +13 -10
- data/lib/bio-ucsc/taegut1/mrna.rb +13 -10
- data/lib/bio-ucsc/taegut1/rmsk.rb +12 -9
- data/lib/bio-ucsc/tetnig2.rb +6 -4
- data/lib/bio-ucsc/tetnig2/mrna.rb +11 -8
- data/lib/bio-ucsc/uniprot.rb +7 -4
- data/lib/bio-ucsc/visigene.rb +6 -3
- data/lib/bio-ucsc/xentro2.rb +6 -12
- data/samples/num-gene-exon.rb +1 -2
- data/samples/snp2genes.rb +1 -7
- data/samples/symbol2summary.rb +7 -1
- data/spec/ailmel1_spec.rb +245 -245
- data/spec/aplcal1_spec.rb +0 -30
- data/spec/braflo1_spec.rb +1 -1
- data/spec/canfam2_spec.rb +6 -6
- data/spec/danrer7_spec.rb +6 -6
- data/spec/file/twobit.rb +167 -0
- data/spec/galgal3_spec.rb +12 -12
- data/spec/gasacu1_spec.rb +6 -6
- data/spec/hg18/find_by_spec.rb +14 -20
- data/spec/hg19/all_mrna_spec.rb +0 -8
- data/spec/hg19/altseqliftoverpsl_spec.rb +0 -8
- data/spec/hg19/altseqpatches_spec.rb +0 -8
- data/spec/hg19/{chainanocar1_spec.rb → chainanocar2_spec.rb} +2 -2
- data/spec/hg19/{chainbostau4_spec.rb → chainbostau6_spec.rb} +2 -2
- data/spec/hg19/{chainxentro2_spec.rb → chainxentro3_spec.rb} +2 -2
- data/spec/hg19/connect_spec.rb +20 -0
- data/spec/hg19/find_by_spec.rb +14 -14
- data/spec/hg19/gwascatalog_spec.rb +0 -7
- data/spec/hg19/hgikmcextra_spec.rb +2 -1
- data/spec/hg19/mrnaorinetinfo_spec.rb +0 -8
- data/spec/hg19/{netanocar1_spec.rb → netanocar2_spec.rb} +2 -2
- data/spec/hg19/{netbostau4_spec.rb → netbostau6_spec.rb} +2 -2
- data/spec/hg19/{netxentro2_spec.rb → netxentro3_spec.rb} +2 -2
- data/spec/hg19/sibtxgraph_spec.rb +0 -8
- data/spec/hg19/snp132_spec.rb +9 -0
- data/spec/hg19/xenorefflat_spec.rb +0 -8
- data/spec/hg19/xenorefgene_spec.rb +0 -8
- data/spec/hg19/xenorefseqali_spec.rb +0 -8
- data/spec/mm9_spec.rb +32 -32
- data/spec/ornana1_spec.rb +12 -12
- data/spec/{named_scope_spec.rb → relation_objects_spec.rb} +14 -0
- data/spec/rhemac2_spec.rb +16 -6
- data/spec/rn4_spec.rb +12 -12
- data/spec/spec_helper.rb +3 -1
- data/spec/susscr2_spec.rb +6 -6
- data/spec/taegut1_spec.rb +8 -8
- data/spec/xentro2_spec.rb +6 -6
- metadata +63 -138
- data/README.rdoc +0 -202
- data/lib/bio-ucsc/ailmel1/db_connection.rb +0 -57
- data/lib/bio-ucsc/anocar2/db_connection.rb +0 -57
- data/lib/bio-ucsc/anogam1/db_connection.rb +0 -57
- data/lib/bio-ucsc/aplcal1/db_connection.rb +0 -57
- data/lib/bio-ucsc/bostau4/db_connection.rb +0 -57
- data/lib/bio-ucsc/braflo1/db_connection.rb +0 -57
- data/lib/bio-ucsc/caejap1/db_connection.rb +0 -57
- data/lib/bio-ucsc/caepb2/db_connection.rb +0 -57
- data/lib/bio-ucsc/caerem3/db_connection.rb +0 -57
- data/lib/bio-ucsc/caljac3/db_connection.rb +0 -55
- data/lib/bio-ucsc/canfam2/db_connection.rb +0 -57
- data/lib/bio-ucsc/cavpor3/db_connection.rb +0 -57
- data/lib/bio-ucsc/cb3/db_connection.rb +0 -57
- data/lib/bio-ucsc/ce6/db_connection.rb +0 -55
- data/lib/bio-ucsc/ci2/db_connection.rb +0 -55
- data/lib/bio-ucsc/danrer7/db_connection.rb +0 -57
- data/lib/bio-ucsc/dm3/db_connection.rb +0 -55
- data/lib/bio-ucsc/dp3/db_connection.rb +0 -57
- data/lib/bio-ucsc/droana2/db_connection.rb +0 -57
- data/lib/bio-ucsc/droere1/db_connection.rb +0 -57
- data/lib/bio-ucsc/drogri1/db_connection.rb +0 -57
- data/lib/bio-ucsc/dromoj2/db_connection.rb +0 -57
- data/lib/bio-ucsc/droper1/db_connection.rb +0 -57
- data/lib/bio-ucsc/drosec1/db_connection.rb +0 -57
- data/lib/bio-ucsc/drosim1/db_connection.rb +0 -57
- data/lib/bio-ucsc/drovir2/db_connection.rb +0 -57
- data/lib/bio-ucsc/droyak2/db_connection.rb +0 -57
- data/lib/bio-ucsc/equcab2/db_connection.rb +0 -57
- data/lib/bio-ucsc/felcat4/db_connection.rb +0 -57
- data/lib/bio-ucsc/fr2/db_connection.rb +0 -57
- data/lib/bio-ucsc/galgal3/db_connection.rb +0 -57
- data/lib/bio-ucsc/gasacu1/db_connection.rb +0 -57
- data/lib/bio-ucsc/go/db_connection.rb +0 -55
- data/lib/bio-ucsc/hg18/db_connection.rb +0 -59
- data/lib/bio-ucsc/hg19/db_connection.rb +0 -61
- data/lib/bio-ucsc/hgfixed/db_connection.rb +0 -55
- data/lib/bio-ucsc/loxafr3/db_connection.rb +0 -57
- data/lib/bio-ucsc/mm9/chainanocar1.rb +0 -79
- data/lib/bio-ucsc/mm9/chainanocar1link.rb +0 -79
- data/lib/bio-ucsc/mm9/chainbostau4.rb +0 -79
- data/lib/bio-ucsc/mm9/chainbostau4link.rb +0 -79
- data/lib/bio-ucsc/mm9/chainpantro2.rb +0 -81
- data/lib/bio-ucsc/mm9/chainpantro2link.rb +0 -82
- data/lib/bio-ucsc/mm9/chainxentro2.rb +0 -81
- data/lib/bio-ucsc/mm9/chainxentro2link.rb +0 -81
- data/lib/bio-ucsc/mm9/db_connection.rb +0 -55
- data/lib/bio-ucsc/mondom5/db_connection.rb +0 -57
- data/lib/bio-ucsc/ornana1/db_connection.rb +0 -57
- data/lib/bio-ucsc/orycun2/db_connection.rb +0 -57
- data/lib/bio-ucsc/orylat2/db_connection.rb +0 -55
- data/lib/bio-ucsc/oviari1/db_connection.rb +0 -57
- data/lib/bio-ucsc/pantro3/db_connection.rb +0 -55
- data/lib/bio-ucsc/petmar1/db_connection.rb +0 -57
- data/lib/bio-ucsc/ponabe2/db_connection.rb +0 -55
- data/lib/bio-ucsc/pripac1/db_connection.rb +0 -57
- data/lib/bio-ucsc/proteome/db_connection.rb +0 -59
- data/lib/bio-ucsc/rhemac2/chainpantro2.rb +0 -81
- data/lib/bio-ucsc/rhemac2/chainpantro2link.rb +0 -81
- data/lib/bio-ucsc/rhemac2/db_connection.rb +0 -55
- data/lib/bio-ucsc/rn4/chainbostau3.rb +0 -79
- data/lib/bio-ucsc/rn4/chainbostau3link.rb +0 -79
- data/lib/bio-ucsc/rn4/chainxentro2.rb +0 -81
- data/lib/bio-ucsc/rn4/chainxentro2link.rb +0 -81
- data/lib/bio-ucsc/rn4/db_connection.rb +0 -55
- data/lib/bio-ucsc/saccer2/db_connection.rb +0 -55
- data/lib/bio-ucsc/strpur2/chaingalgal3.rb +0 -81
- data/lib/bio-ucsc/strpur2/chaingalgal3link.rb +0 -81
- data/lib/bio-ucsc/strpur2/chainhg18.rb +0 -81
- data/lib/bio-ucsc/strpur2/chainhg18link.rb +0 -81
- data/lib/bio-ucsc/strpur2/chainmm9.rb +0 -81
- data/lib/bio-ucsc/strpur2/chainmm9link.rb +0 -81
- data/lib/bio-ucsc/strpur2/chainpetmar1.rb +0 -81
- data/lib/bio-ucsc/strpur2/chainpetmar1link.rb +0 -81
- data/lib/bio-ucsc/strpur2/db_connection.rb +0 -57
- data/lib/bio-ucsc/strpur2/gap.rb +0 -81
- data/lib/bio-ucsc/strpur2/gold.rb +0 -81
- data/lib/bio-ucsc/susscr2/db_connection.rb +0 -57
- data/lib/bio-ucsc/taegut1/db_connection.rb +0 -57
- data/lib/bio-ucsc/tetnig2/db_connection.rb +0 -57
- data/lib/bio-ucsc/uniprot/db_connection.rb +0 -59
- data/lib/bio-ucsc/visigene/db_connection.rb +0 -55
- data/lib/bio-ucsc/xentro2/chaingalgal3.rb +0 -81
- data/lib/bio-ucsc/xentro2/chaingalgal3link.rb +0 -81
- data/lib/bio-ucsc/xentro2/db_connection.rb +0 -57
- data/lib/bio-ucsc/xentro2/est.rb +0 -81
- data/lib/bio-ucsc/xentro2/gap.rb +0 -81
- data/lib/bio-ucsc/xentro2/gold.rb +0 -81
- data/lib/bio-ucsc/xentro2/intronest.rb +0 -81
- data/lib/bio-ucsc/xentro2/mrna.rb +0 -81
- data/lib/bio-ucsc/xentro2/rmsk.rb +0 -78
- data/spec/hg19/chainaplcal1_spec.rb +0 -16
- data/spec/hg19/netaplcal1_spec.rb +0 -16
@@ -13,21 +13,24 @@ module Bio
|
|
13
13
|
module EquCab2
|
14
14
|
|
15
15
|
class Rmsk
|
16
|
+
include DBConnector
|
17
|
+
DBConnection.database "equCab2"
|
18
|
+
|
16
19
|
Bio::Ucsc::EquCab2::CHROMS.each do |chr|
|
17
20
|
class_eval %!
|
18
21
|
class #{chr[0..0].upcase + chr[1..-1]}_Rmsk < DBConnection
|
19
|
-
|
20
|
-
|
22
|
+
self.table_name = "#{chr[0..0].downcase + chr[1..-1]}_rmsk"
|
23
|
+
self.primary_key = nil
|
21
24
|
|
22
|
-
def self.find_by_interval(interval, opt = {:partial => true})
|
25
|
+
def self.find_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
23
26
|
find_first_or_all_by_interval(interval, :first, opt)
|
24
27
|
end
|
25
28
|
|
26
|
-
def self.find_all_by_interval(interval, opt = {:partial => true})
|
29
|
+
def self.find_all_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
27
30
|
find_first_or_all_by_interval(interval, :all, opt)
|
28
31
|
end
|
29
32
|
|
30
|
-
def self.find_first_or_all_by_interval(interval, first_all, opt)
|
33
|
+
def self.find_first_or_all_by_interval(interval, first_all, opt); interval = Bio::Ucsc::Gi.wrap(interval)
|
31
34
|
zstart = interval.zero_start
|
32
35
|
zend = interval.zero_end
|
33
36
|
if opt[:partial] == true
|
@@ -60,13 +63,13 @@ AND (genoEnd BETWEEN :zstart AND :zend))
|
|
60
63
|
!
|
61
64
|
end # each chromosome
|
62
65
|
|
63
|
-
def self.find_by_interval(interval, opt = {:partial => true})
|
66
|
+
def self.find_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
64
67
|
chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
|
65
68
|
chr_klass = self.const_get("#{chrom}_Rmsk")
|
66
69
|
chr_klass.__send__(:find_by_interval, interval, opt)
|
67
70
|
end
|
68
71
|
|
69
|
-
def self.find_all_by_interval(interval, opt = {:partial => true})
|
72
|
+
def self.find_all_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
70
73
|
chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
|
71
74
|
chr_klass = self.const_get("#{chrom}_Rmsk")
|
72
75
|
chr_klass.__send__(:find_all_by_interval, interval, opt)
|
data/lib/bio-ucsc/felcat4.rb
CHANGED
@@ -6,8 +6,8 @@
|
|
6
6
|
# License:: The Ruby licence (Ryby's / GPLv2 dual)
|
7
7
|
#
|
8
8
|
|
9
|
-
require "#{File.dirname(__FILE__)}/
|
10
|
-
require "#{File.dirname(__FILE__)}/table_class_detector"
|
9
|
+
require "#{::File.dirname(__FILE__)}/db_connector"
|
10
|
+
require "#{::File.dirname(__FILE__)}/table_class_detector"
|
11
11
|
|
12
12
|
module Bio
|
13
13
|
module Ucsc
|
@@ -15,9 +15,12 @@ module Bio
|
|
15
15
|
CHROMS =
|
16
16
|
%w(chrA1 chrC1 chrA2 chrB1 chrA3 chrC2 chrB2 chrB4 chrB3
|
17
17
|
chrD1 chrX chrE1 chrD3 chrD4 chrD2 chrF1 chrF2 chrE2 chrE3)
|
18
|
-
extend TableClassDetector
|
19
18
|
|
20
|
-
|
19
|
+
extend TableClassDetector
|
20
|
+
include DBConnector
|
21
|
+
DBConnection.database "felCat4"
|
22
|
+
|
23
|
+
base = "#{::File.dirname(__FILE__)}/felcat4"
|
21
24
|
end
|
22
25
|
end
|
23
26
|
end
|
@@ -0,0 +1,10 @@
|
|
1
|
+
#
|
2
|
+
# = the Bio::Ucsc::File module
|
3
|
+
# Copyright:: Copyright (C) 2011-2012
|
4
|
+
# MISHIMA, Hiroyuki
|
5
|
+
# <missy at be.to / hmishima at nagasaki-u.ac.jp>
|
6
|
+
# License:: The Ruby licence (Ryby's / GPLv2 dual)
|
7
|
+
#
|
8
|
+
|
9
|
+
require "#{::File.dirname(__FILE__)}/file/twobit"
|
10
|
+
require "#{::File.dirname(__FILE__)}/file/bytequeue"
|
@@ -0,0 +1,28 @@
|
|
1
|
+
#
|
2
|
+
# = bytequeue.rb
|
3
|
+
# a class for byte stream queues
|
4
|
+
#
|
5
|
+
# Copyright:: Copyright (C) 2011, 2012
|
6
|
+
# MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp>
|
7
|
+
# License:: Ruby license (Ryby's / GPLv2 dual)
|
8
|
+
|
9
|
+
module Bio
|
10
|
+
module Ucsc
|
11
|
+
module File
|
12
|
+
class ByteQueue
|
13
|
+
def initialize(str)
|
14
|
+
@str = str
|
15
|
+
@index = 0
|
16
|
+
end
|
17
|
+
|
18
|
+
attr_accessor :index
|
19
|
+
|
20
|
+
def next(n)
|
21
|
+
result = @str[@index, n]
|
22
|
+
@index += n
|
23
|
+
result
|
24
|
+
end
|
25
|
+
end # class ByteQueue
|
26
|
+
end # module File
|
27
|
+
end # module Ucsc
|
28
|
+
end # module Bio
|
@@ -0,0 +1,174 @@
|
|
1
|
+
#
|
2
|
+
# = twobit.rb
|
3
|
+
# handle UCSC's 2bit file (locally stored) to retrieve the reference sequence
|
4
|
+
#
|
5
|
+
# Copyright:: Copyright (C) 2011, 2012
|
6
|
+
# MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp>
|
7
|
+
# License:: Ruby license (Ryby's / GPLv2 dual)
|
8
|
+
|
9
|
+
module Bio
|
10
|
+
module Ucsc
|
11
|
+
module File
|
12
|
+
class Twobit
|
13
|
+
TwoBitHeader =
|
14
|
+
Struct.new(:signature, :version, :sequence_count, :reserved)
|
15
|
+
TwoBitRecord =
|
16
|
+
Struct.new(:dna_size,
|
17
|
+
:n_block_intervals, :mask_block_intervals,
|
18
|
+
:reserved, :packed_dna_offset)
|
19
|
+
BINCODE = {0b00 => "T", 0b01 => "C", 0b10 => "A", 0b11 => "G"}
|
20
|
+
|
21
|
+
def initialize(twobit_queue, filename, header, offsets)
|
22
|
+
@tbq = twobit_queue
|
23
|
+
@filename = filename
|
24
|
+
@header = header
|
25
|
+
@offsets = offsets
|
26
|
+
@records = Hash.new
|
27
|
+
end
|
28
|
+
|
29
|
+
attr_reader :filename, :header, :offsets
|
30
|
+
|
31
|
+
def self.load(filename)
|
32
|
+
two_bit = nil
|
33
|
+
Kernel.open(filename, 'rb') {|f| two_bit = f.read}
|
34
|
+
tbq = Bio::Ucsc::File::ByteQueue.new(two_bit)
|
35
|
+
|
36
|
+
twobit_header = TwoBitHeader.new
|
37
|
+
twobit_header.signature = tbq.next(4).unpack('L').first
|
38
|
+
twobit_header.version = tbq.next(4).unpack('L').first
|
39
|
+
twobit_header.sequence_count = tbq.next(4).unpack('L').first
|
40
|
+
twobit_header.reserved = tbq.next(4).unpack('L').first
|
41
|
+
|
42
|
+
offsets = Hash.new
|
43
|
+
twobit_header.sequence_count.times do
|
44
|
+
name_length = tbq.next(1).unpack('C').first
|
45
|
+
offsets[tbq.next(name_length).unpack('a*').first] =
|
46
|
+
tbq.next(4).unpack('L').first
|
47
|
+
end
|
48
|
+
new(tbq, filename, twobit_header, offsets)
|
49
|
+
end
|
50
|
+
|
51
|
+
def self.open(filename)
|
52
|
+
if block_given?
|
53
|
+
yield self.load(filename)
|
54
|
+
else
|
55
|
+
return self.load(filename)
|
56
|
+
end
|
57
|
+
end
|
58
|
+
|
59
|
+
def records(chrom)
|
60
|
+
return @records[chrom] if @records[chrom]
|
61
|
+
|
62
|
+
@tbq.index = @offsets[chrom]
|
63
|
+
@records[chrom] = TwoBitRecord.new
|
64
|
+
@records[chrom].dna_size = @tbq.next(4).unpack('L').first
|
65
|
+
|
66
|
+
n_block_count = @tbq.next(4).unpack('L').first
|
67
|
+
n_block_starts = Array.new
|
68
|
+
n_block_count.times do
|
69
|
+
n_block_starts << @tbq.next(4).unpack('L').first
|
70
|
+
end
|
71
|
+
n_block_sizes = Array.new
|
72
|
+
n_block_count.times do
|
73
|
+
n_block_sizes << @tbq.next(4).unpack('L').first
|
74
|
+
end
|
75
|
+
@records[chrom].n_block_intervals = Array.new
|
76
|
+
n_block_count.times do |idx|
|
77
|
+
@records[chrom].n_block_intervals <<
|
78
|
+
Bio::GenomicInterval.zero_based(chrom,
|
79
|
+
n_block_starts[idx],
|
80
|
+
n_block_starts[idx]+n_block_sizes[idx])
|
81
|
+
end
|
82
|
+
|
83
|
+
mask_block_count = @tbq.next(4).unpack('L').first
|
84
|
+
mask_block_starts = Array.new
|
85
|
+
mask_block_count.times do
|
86
|
+
mask_block_starts << @tbq.next(4).unpack('L').first
|
87
|
+
end
|
88
|
+
mask_block_sizes = Array.new
|
89
|
+
mask_block_count.times do
|
90
|
+
mask_block_sizes << @tbq.next(4).unpack('L').first
|
91
|
+
end
|
92
|
+
@records[chrom].mask_block_intervals = Array.new
|
93
|
+
mask_block_count.times do |idx|
|
94
|
+
@records[chrom].mask_block_intervals <<
|
95
|
+
Bio::GenomicInterval.zero_based(chrom,
|
96
|
+
mask_block_starts[idx],
|
97
|
+
mask_block_starts[idx]+mask_block_sizes[idx])
|
98
|
+
end
|
99
|
+
|
100
|
+
@records[chrom].reserved = @tbq.next(4).unpack('L').first
|
101
|
+
@records[chrom].packed_dna_offset = @tbq.index
|
102
|
+
|
103
|
+
@records[chrom]
|
104
|
+
end
|
105
|
+
|
106
|
+
def subseq(interval)
|
107
|
+
find_by_interval(interval)
|
108
|
+
end
|
109
|
+
|
110
|
+
def find_by_interval(interval)
|
111
|
+
interval = Bio::Ucsc::Gi.wrap(interval)
|
112
|
+
seq = self.find_by_interval_raw(interval)
|
113
|
+
@records[interval.chrom].n_block_intervals.map do |nb|
|
114
|
+
if interval.overlapped?(nb)
|
115
|
+
case interval.compare(nb)
|
116
|
+
when :equal,:contained_by
|
117
|
+
seq = 'N' * interval.overlap(nb)
|
118
|
+
when :contains
|
119
|
+
left_len = nb.chr_start - interval.chr_start + 1
|
120
|
+
right_len = interval.chr_end - nb.chr_end + 1
|
121
|
+
seq[0, left_len] = 'N' * left_len
|
122
|
+
seq[-right_len, right_len] = 'N' * right_len
|
123
|
+
when :left_overlapped
|
124
|
+
left_len = nb.chr_end - interval.chr_start + 1
|
125
|
+
seq[0, left_len] = 'N' * left_len
|
126
|
+
when :right_overlapped
|
127
|
+
right_len = interval.chr_end - nb.chr_start + 1
|
128
|
+
seq[-right_len, right_len] = 'N' * right_len
|
129
|
+
when :right_adjacent, :right_off
|
130
|
+
# expecting that N-blocks are sorted
|
131
|
+
# return Bio::Sequence::NA.new(seq)
|
132
|
+
seq
|
133
|
+
end
|
134
|
+
end
|
135
|
+
end
|
136
|
+
#Bio::Sequence::NA.new(seq)
|
137
|
+
seq
|
138
|
+
end
|
139
|
+
|
140
|
+
def find_by_interval_raw(interval)
|
141
|
+
interval = Bio::Ucsc::Gi.wrap(interval)
|
142
|
+
byte_count, byte_mod = interval.zero_start.divmod 4
|
143
|
+
chrom_top = self.records(interval.chrom).packed_dna_offset
|
144
|
+
div_start, mod_start = interval.zero_start.divmod 4
|
145
|
+
div_end, mod_end = interval.zero_end.divmod 4
|
146
|
+
div_len, mod_len = interval.length.divmod 4
|
147
|
+
|
148
|
+
byte_length = div_end - div_start + 1
|
149
|
+
@tbq.index = chrom_top + div_start
|
150
|
+
bytes = @tbq.next(byte_length).unpack('C*')
|
151
|
+
seq = bytes_to_nucleotides(bytes)
|
152
|
+
seq[mod_start..(-1-(4-mod_end))]
|
153
|
+
end
|
154
|
+
|
155
|
+
private
|
156
|
+
|
157
|
+
def bytes_to_nucleotides(bytes)
|
158
|
+
results = ""
|
159
|
+
bytes.each do |byte|
|
160
|
+
results << byte_to_nucleotides(byte)
|
161
|
+
end
|
162
|
+
results
|
163
|
+
end
|
164
|
+
|
165
|
+
def byte_to_nucleotides(byte)
|
166
|
+
BINCODE[byte >> 6] +
|
167
|
+
BINCODE[(byte >> 4) & 0b11] +
|
168
|
+
BINCODE[(byte >> 2) & 0b11] +
|
169
|
+
BINCODE[byte & 0b11]
|
170
|
+
end
|
171
|
+
end # class ReferenceSequence
|
172
|
+
end # module File
|
173
|
+
end # module Ucsc
|
174
|
+
end # module Bio
|
data/lib/bio-ucsc/fr2.rb
CHANGED
@@ -6,16 +6,18 @@
|
|
6
6
|
# License:: The Ruby licence (Ryby's / GPLv2 dual)
|
7
7
|
#
|
8
8
|
|
9
|
-
require "#{File.dirname(__FILE__)}/
|
10
|
-
require "#{File.dirname(__FILE__)}/table_class_detector"
|
9
|
+
require "#{::File.dirname(__FILE__)}/db_connector"
|
10
|
+
require "#{::File.dirname(__FILE__)}/table_class_detector"
|
11
11
|
|
12
12
|
module Bio
|
13
13
|
module Ucsc
|
14
14
|
module Fr2
|
15
15
|
CHROMS = %w(ChrUn ChrM)
|
16
16
|
extend TableClassDetector
|
17
|
+
include DBConnector
|
18
|
+
DBConnection.database "fr2"
|
17
19
|
|
18
|
-
base = "#{File.dirname(__FILE__)}/fr2"
|
20
|
+
base = "#{::File.dirname(__FILE__)}/fr2"
|
19
21
|
autoload :ChainGalGal3, "#{base}/chaingalgal3"
|
20
22
|
autoload :ChainGalGal3Link, "#{base}/chaingalgal3link"
|
21
23
|
autoload :ChainGasAcu1, "#{base}/chaingasacu1"
|
@@ -12,25 +12,28 @@ module Bio
|
|
12
12
|
module Fr2
|
13
13
|
|
14
14
|
class ChainGalGal3
|
15
|
+
include DBConnector
|
16
|
+
DBConnection.database "fr2"
|
17
|
+
|
15
18
|
KLASS = "ChainGalGal3"
|
16
19
|
KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
|
17
20
|
|
18
21
|
Bio::Ucsc::Fr2::CHROMS.each do |chr|
|
19
22
|
class_eval %!
|
20
23
|
class #{chr[0..0].upcase + chr[1..-1]}_#{KLASS} < DBConnection
|
21
|
-
|
22
|
-
|
23
|
-
|
24
|
+
self.table_name = "#{chr[0..0].downcase + chr[1..-1]}_#{KLASS_S}"
|
25
|
+
self.primary_key = nil
|
26
|
+
self.inheritance_column = nil
|
24
27
|
|
25
|
-
def self.find_by_interval(interval, opt = {:partial => true})
|
28
|
+
def self.find_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
26
29
|
find_first_or_all_by_interval(interval, :first, opt)
|
27
30
|
end
|
28
31
|
|
29
|
-
def self.find_all_by_interval(interval, opt = {:partial => true})
|
32
|
+
def self.find_all_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
30
33
|
find_first_or_all_by_interval(interval, :all, opt)
|
31
34
|
end
|
32
35
|
|
33
|
-
def self.find_first_or_all_by_interval(interval, first_all, opt)
|
36
|
+
def self.find_first_or_all_by_interval(interval, first_all, opt); interval = Bio::Ucsc::Gi.wrap(interval)
|
34
37
|
zstart = interval.zero_start
|
35
38
|
zend = interval.zero_end
|
36
39
|
if opt[:partial] == true
|
@@ -63,13 +66,13 @@ AND (tEnd BETWEEN :zstart AND :zend))
|
|
63
66
|
!
|
64
67
|
end # each chromosome
|
65
68
|
|
66
|
-
def self.find_by_interval(interval, opt = {:partial => true})
|
69
|
+
def self.find_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
67
70
|
chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
|
68
71
|
chr_klass = self.const_get("#{chrom}_#{KLASS}")
|
69
72
|
chr_klass.__send__(:find_by_interval, interval, opt)
|
70
73
|
end
|
71
74
|
|
72
|
-
def self.find_all_by_interval(interval, opt = {:partial => true})
|
75
|
+
def self.find_all_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
73
76
|
chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
|
74
77
|
chr_klass = self.const_get("#{chrom}_#{KLASS}")
|
75
78
|
chr_klass.__send__(:find_all_by_interval, interval, opt)
|
@@ -12,25 +12,28 @@ module Bio
|
|
12
12
|
module Fr2
|
13
13
|
|
14
14
|
class ChainGalGal3Link
|
15
|
+
include DBConnector
|
16
|
+
DBConnection.database "fr2"
|
17
|
+
|
15
18
|
KLASS = "ChainGalGal3Link"
|
16
19
|
KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
|
17
20
|
|
18
21
|
Bio::Ucsc::Fr2::CHROMS.each do |chr|
|
19
22
|
class_eval %!
|
20
23
|
class #{chr[0..0].upcase + chr[1..-1]}_#{KLASS} < DBConnection
|
21
|
-
|
22
|
-
|
23
|
-
|
24
|
+
self.table_name = "#{chr[0..0].downcase + chr[1..-1]}_#{KLASS_S}"
|
25
|
+
self.primary_key = nil
|
26
|
+
self.inheritance_column = nil
|
24
27
|
|
25
|
-
def self.find_by_interval(interval, opt = {:partial => true})
|
28
|
+
def self.find_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
26
29
|
find_first_or_all_by_interval(interval, :first, opt)
|
27
30
|
end
|
28
31
|
|
29
|
-
def self.find_all_by_interval(interval, opt = {:partial => true})
|
32
|
+
def self.find_all_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
30
33
|
find_first_or_all_by_interval(interval, :all, opt)
|
31
34
|
end
|
32
35
|
|
33
|
-
def self.find_first_or_all_by_interval(interval, first_all, opt)
|
36
|
+
def self.find_first_or_all_by_interval(interval, first_all, opt); interval = Bio::Ucsc::Gi.wrap(interval)
|
34
37
|
zstart = interval.zero_start
|
35
38
|
zend = interval.zero_end
|
36
39
|
if opt[:partial] == true
|
@@ -63,13 +66,13 @@ AND (tEnd BETWEEN :zstart AND :zend))
|
|
63
66
|
!
|
64
67
|
end # each chromosome
|
65
68
|
|
66
|
-
def self.find_by_interval(interval, opt = {:partial => true})
|
69
|
+
def self.find_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
67
70
|
chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
|
68
71
|
chr_klass = self.const_get("#{chrom}_#{KLASS}")
|
69
72
|
chr_klass.__send__(:find_by_interval, interval, opt)
|
70
73
|
end
|
71
74
|
|
72
|
-
def self.find_all_by_interval(interval, opt = {:partial => true})
|
75
|
+
def self.find_all_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
73
76
|
chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
|
74
77
|
chr_klass = self.const_get("#{chrom}_#{KLASS}")
|
75
78
|
chr_klass.__send__(:find_all_by_interval, interval, opt)
|
@@ -12,25 +12,28 @@ module Bio
|
|
12
12
|
module Fr2
|
13
13
|
|
14
14
|
class ChainGasAcu1
|
15
|
+
include DBConnector
|
16
|
+
DBConnection.database "fr2"
|
17
|
+
|
15
18
|
KLASS = "ChainGasAcu1"
|
16
19
|
KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
|
17
20
|
|
18
21
|
Bio::Ucsc::Fr2::CHROMS.each do |chr|
|
19
22
|
class_eval %!
|
20
23
|
class #{chr[0..0].upcase + chr[1..-1]}_#{KLASS} < DBConnection
|
21
|
-
|
22
|
-
|
23
|
-
|
24
|
+
self.table_name = "#{chr[0..0].downcase + chr[1..-1]}_#{KLASS_S}"
|
25
|
+
self.primary_key = nil
|
26
|
+
self.inheritance_column = nil
|
24
27
|
|
25
|
-
def self.find_by_interval(interval, opt = {:partial => true})
|
28
|
+
def self.find_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
26
29
|
find_first_or_all_by_interval(interval, :first, opt)
|
27
30
|
end
|
28
31
|
|
29
|
-
def self.find_all_by_interval(interval, opt = {:partial => true})
|
32
|
+
def self.find_all_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
30
33
|
find_first_or_all_by_interval(interval, :all, opt)
|
31
34
|
end
|
32
35
|
|
33
|
-
def self.find_first_or_all_by_interval(interval, first_all, opt)
|
36
|
+
def self.find_first_or_all_by_interval(interval, first_all, opt); interval = Bio::Ucsc::Gi.wrap(interval)
|
34
37
|
zstart = interval.zero_start
|
35
38
|
zend = interval.zero_end
|
36
39
|
if opt[:partial] == true
|
@@ -63,13 +66,13 @@ AND (tEnd BETWEEN :zstart AND :zend))
|
|
63
66
|
!
|
64
67
|
end # each chromosome
|
65
68
|
|
66
|
-
def self.find_by_interval(interval, opt = {:partial => true})
|
69
|
+
def self.find_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
67
70
|
chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
|
68
71
|
chr_klass = self.const_get("#{chrom}_#{KLASS}")
|
69
72
|
chr_klass.__send__(:find_by_interval, interval, opt)
|
70
73
|
end
|
71
74
|
|
72
|
-
def self.find_all_by_interval(interval, opt = {:partial => true})
|
75
|
+
def self.find_all_by_interval(interval, opt = {:partial => true}); interval = Bio::Ucsc::Gi.wrap(interval)
|
73
76
|
chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
|
74
77
|
chr_klass = self.const_get("#{chrom}_#{KLASS}")
|
75
78
|
chr_klass.__send__(:find_all_by_interval, interval, opt)
|