def mm9_17():
dm.deriveNewDataset("mm9.17.Imprinted_DMR.alleleSpecific_differentiallyMethylatedRegions_manuallyCurated.region",
"mm9_alleleSpecificDMR",
[],
[],
None,
None,
True)
def mm9_12():
dm.deriveNewDataset("mm9.12.Enhancer_p300.ChIPseq_p300_binding_sites_Visel2009_noPromoter.region",
"mm9_enhancers_Visel_all",
["mm9.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region"],
[],
"max",
None,
True)
def mm9_19():
dm.deriveNewDataset("mm9.19.CancerGene_Promoter.Promoter_region_ensemblGenesMinus10000ToPlus2000AroundTSS_cancerGeneCensus.region",
"mm9_CancerGene_Promoters",
[],
[],
None,
None,
True)
def mm9_7():
dm.deriveNewDataset("mm9.7.NonCGI_Promoter.Promoter_centered_nonOverlapping_with_BonaFide_CGI.region",
"mm9.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region",
["mm9.5.BonaFide_CGI.cgiHunter_oe0.6_gc50_len700.region"],
[],
None,
None,
True)
def mm9_5():
dm.deriveNewDataset("mm9.5.BonaFide_CGI.cgiHunter_oe0.6_gc50_len700.region",
"mm9_CGI_CGIHunter_GG",
[],
[],
None,
700,
True)
def mm9_4():
dm.deriveNewDataset("mm9.4.CpG_Island.cgiHunter_oe0.6_gc50_len200.region",
"mm9_CGI_CGIHunter_GG",
[],
[],
None,
200,
True)
def mm9_20():
dm.deriveNewDataset("mm9.20.CellStateGene_Promoter.Promoter_region_ensemblGenesMinus10000ToPlus2000AroundTSS_manuallyCurated.region",
"mm9_CandidatePromoters",
[],
[],
None,
None,
True)
def hg18_13():
dm.deriveNewDataset("hg18.13.Enhancer_H3K4me2.ChIPseq_H3K4me2_sites_Heintzman2009_noPromoter.region",
"hg18_enhancers_Heintzman2009",
["hg18.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region"],
[],
None,
None,
True)
def hg18_12():
dm.deriveNewDataset("hg18_support.12.DNaseI_HS_site.hg18_tCell_DNaseI_hypersensitive_sites_Boyle2008",
"hg18_DNAseCD4Sites",
[],
[],
None,
None,
True)
dm.regionDatasets.update(regionDatasets_DNaseI_Duke)
dm.mergeDatasets("hg18_support.12.Duke_GM12878_HelaS3_K562_hepg2_SortedNotMerged",regionDatasets_DNaseI_Duke.keys())
dm.deriveNewDataset("hg18_support.12.Duke_GM12878_HelaS3_K562_hepg2",
"hg18_support.12.Duke_GM12878_HelaS3_K562_hepg2_SortedNotMerged",
[],
[],
"max",
None,
True,
True)
#
dm.regionDatasets.update(regionDatasets_DNaseI_Washington)
dm.mergeDatasets("hg18_support.12.Open_Chromatin.ENCODE_DNaseI_hypersensitive_sites_SortedNotMerged",
regionDatasets_DNaseI_Washington.keys())
dm.deriveNewDataset("hg18_support.12.Open_Chromatin.ENCODE_DNaseI_hypersensitive_sites",
"hg18_support.12.Open_Chromatin.ENCODE_DNaseI_hypersensitive_sites_SortedNotMerged",
[],
[],
"max",
None,
True,
True)
dm.mergeDatasets("hg18_support.12.All_DNaseI_datasets_merged",
["hg18_support.12.Open_Chromatin.ENCODE_DNaseI_hypersensitive_sites",
"hg18_support.12.Duke_GM12878_HelaS3_K562_hepg2",
"hg18_support.12.DNaseI_HS_site.hg18_tCell_DNaseI_hypersensitive_sites_Boyle2008"
])
dm.deriveNewDataset("hg18.12.Open_Chromatin.DNase_FAIRE_peaks_ENCODE_noPromoter.region",
"hg18_support.12.All_DNaseI_datasets_merged",
["hg18.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region"],
[],
"max",
None,
True,
True)
dm.deriveNewDataset("hg18.12.Open_Chromatin.DNase_FAIRE_peaks_ENCODE_noPromoter_downsampledTo500k.region",
"hg18.12.Open_Chromatin.DNase_FAIRE_peaks_ENCODE_noPromoter.region",
[],
[],
None,
None,
True,
True,
500000)
def mm9_13():
dm.deriveNewDataset(
"mm9.13.Enhancer_H3K4me2.ChIPseq_H3K4me2_sitesWithin1kbTo100kbOfPromoter_Meissner2008_noPromoter.region",
"mm9_enhancers_meissner_all",
["mm9.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region"],
[],
"max",
None,
True)
def mm9_22():
dm.regionDatasets.update(regionDatasetsRepeatsMM9)
dm.deriveNewDataset("mm9.22.Repeat_DNA.repeatMasker_annotation_DNA_repeats.region",
"mm9_Repeat_DNA",
[],
[],
None,
None,
True,
True)
def mm9_11():
dm.deriveNewDataset("mm9.11.Conserved_Region.UCSC_mostConserved_minLength100_noPromoter_noCGI_noExon_noUTR.region",
"mm9_mostConserved_all",
["mm9.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region",
"mm9.4.CpG_Island.cgiHunter_oe0.6_gc50_len200.region",
"mm9_RefSeq_allCodingExons","mm9_RefSeq_3UTR"],
[],
"max",
100,
True)
def mm9_9():
dm.deriveNewDataset("mm9.9.Middle_Exon.refSeqGenes_exon_noPromoter_noUTR.region",
"mm9_RefSeq_middleExons",
["mm9.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region",
"mm9_RefSeq_3UTR"],
[],
"max",
None,
True,
True)
def hg18_22():
dm.regionDatasets.update(regionDatasetsRepeatsHG18)
dm.deriveNewDataset("hg18.22.Repeat_DNA.repeatMasker_annotation_DNA_repeats.region",
"hg18_Repeat_DNA",
[],
[],
None,
None,
True,
True)
def hg18_24():
dm.regionDatasets.update(regionDatasetsRepeatsHG18)
dm.deriveNewDataset("hg18.24.Repeat_LowComplex.repeatMasker_annotation_low_complexity_repeats.region",
"hg18_Repeat_LowComplex",
[],
[],
None,
None,
True,
True)
def hg18_26():
dm.regionDatasets.update(regionDatasetsRepeatsHG18)
dm.deriveNewDataset("hg18.26.Repeat_Simple.repeatMasker_annotation_simple_repeats.region",
"hg18_Repeat_Simple",
[],
[],
None,
None,
True,
True)
def mm9_24():
dm.regionDatasets.update(regionDatasetsRepeatsMM9)
dm.deriveNewDataset("mm9.24.Repeat_LowComplex.repeatMasker_annotation_low_complexity_repeats.region",
"mm9_Repeat_LowComplex",
[],
[],
None,
None,
True,
True)
def mm9_8():
dm.extractFlanks("mm9_refseq_genes",
"mm9_support.8.refseq_genes_plus10k",
False,"whole_region",10000,10000)
dm.deriveNewDataset("mm9.8.NonGenic_CGI.cgiHunter_oe0.6_gc50_len200_geneDistance10000.region",
"mm9.4.CpG_Island.cgiHunter_oe0.6_gc50_len200.region",
["mm9_support.8.refseq_genes_plus10k"],
[],
None,
None,
True)
def mm9_10():
dm.deriveNewDataset(
"mm9.10.3prim_UTR.refSeqGenes_3prim_UTR_noPromoter_noExon.region",
"mm9_RefSeq_3UTR",
["mm9.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region",
"mm9_RefSeq_allCodingExons"],
[],
"max",
None,
True,
True)
def mm9_1():
dm.extractFlanks("mm9_ensembl_genes",
"mm9_support.1.Promoter_ensembl.ensemblGenes_minus_5000_to_1000_around_TSS_flancks",
True,"start",5000,1000)
dm.deriveNewDataset("mm9.1.Promoter_ensembl.ensemblGenes_minus_5000_to_1000_around_TSS.region",
"mm9_support.1.Promoter_ensembl.ensemblGenes_minus_5000_to_1000_around_TSS_flancks",
[],
[],
None,
None,
True,
True)
def hg18_3():
dm.extractFlanks("hg18_refseq_genes",
"hg18_support.3.Promoter_region.refSeqGenes_minus_10000_to_2000_around_TSS_flancks",
True,"start",10000,2000)
dm.deriveNewDataset("hg18.3.Promoter_region.refSeqGenes_minus_10000_to_2000_around_TSS.region",
"hg18_support.3.Promoter_region.refSeqGenes_minus_10000_to_2000_around_TSS_flancks",
[],
[],
"max",
None,
True,
True)
def hg18_2():
dm.extractFlanks("hg18_refseq_genes",
"hg18_support.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS_flancks",
True,"start",1000,1000)
dm.deriveNewDataset("hg18.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region",
"hg18_support.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS_flancks",
[],
[],
None,
None,
True,
True)
def hg18_6():
# dm.deriveNewDataset("hg18.6.CGI_Promoter.Promoter_centered_overlapping_with_BonaFide_CGI.region",
# "hg18.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region",
# ["hg18.7.NonCGI_Promoter.Promoter_centered_nonOverlapping_with_BonaFide_CGI.region"],
# [],
# None,
# None,
# True)
dm.deriveNewDataset("hg18.6.CGI_Promoter.Promoter_centered_overlapping_with_BonaFide_CGI.region",
"hg18.2.Promoter_centered.refSeqGenes_minus_1000_to_1000_around_TSS.region",
[],
["hg18.5.BonaFide_CGI.cgiHunter_oe0.6_gc50_len700.region"],
None,
None,
True)
def CancerGenePromoterSufficientMethylationOverlappingWithCGI():
cancerDataset = {
"cancerPromoters_withMethylation":defaultFolder+"mm9.19.CancerGene_Promoter.Promoter_region_ensemblGenesMinus10000ToPlus2000AroundTSS_cancerGeneCensus.region_withMathylation.txt",
}
dm.regionDatasets.update(cancerDataset)
mm9_4()
dm.deriveNewDataset("mm9.cancerPRomoters_CGI",
"cancerPromoters_withMethylation",
[],
["mm9.4.CpG_Island.cgiHunter_oe0.6_gc50_len200.region"],
None,
None,
True,
True)
def mm9_14():
dm.extractFlanks("mm9.4.CpG_Island.cgiHunter_oe0.6_gc50_len200.region",
"mm9_support.14.CGI200_downstream",
False,"start",2000,0)
dm.extractFlanks("mm9.4.CpG_Island.cgiHunter_oe0.6_gc50_len200.region",
"mm9_support.14.CGI200_upstream",
False,"end",0,2000)
dm.mergeDatasets("mm9_support.14.CpG_Island_Shore.cgiHunter_oe0.6_gc50_len200_2kb_outsideCgi",
["mm9_support.14.CGI200_upstream",
"mm9_support.14.CGI200_downstream"])
dm.deriveNewDataset("mm9.14.CpG_Island_Shore.cgiHunter_oe0.6_gc50_len200_2kb_outsideCgi.region",
"mm9_support.14.CpG_Island_Shore.cgiHunter_oe0.6_gc50_len200_2kb_outsideCgi",
[],
[],
"max",
None,
True)
def mm9_15():
dm.extractFlanks("mm9.5.BonaFide_CGI.cgiHunter_oe0.6_gc50_len700.region",
"mm9_support.15.CGI700_downstream",
False,"start",2000,0)
dm.extractFlanks("mm9.5.BonaFide_CGI.cgiHunter_oe0.6_gc50_len700.region",
"mm9_support.15.CGI700_upstream",
False,"end",0,2000)
dm.mergeDatasets("mm9_support.15.BonaFide_CGI_Shore.cgiHunter_oe0.6_gc50_len700_2kb_outsideCgi",
["mm9_support.15.CGI700_downstream",
"mm9_support.15.CGI700_upstream"])
dm.deriveNewDataset("mm9.15.BonaFide_CGI_Shore.cgiHunter_oe0.6_gc50_len700_2kb_outsideCgi.region",
"mm9_support.15.BonaFide_CGI_Shore.cgiHunter_oe0.6_gc50_len700_2kb_outsideCgi",
[],
[],
None,
None,
True)
def hg18_21():
dm.regionDatasets.update(regionDatasetsRepeatsHG18)
dm.deriveNewDataset("hg18.21.Tandem_Repeat.tandemRepeatFinder_annotation.region",
"hg18_tandemRepeat",
[],
[],
None,
None,
True)
dm.deriveNewDataset("hg18.21.Tandem_Repeat.tandemRepeatFinder_annotation_downsampledTo500k.region",
"hg18_tandemRepeat",
[],
[],
None,
None,
True,
True,
500000)
def mm9_21():
dm.regionDatasets.update(regionDatasetsRepeatsMM9)
dm.deriveNewDataset("mm9.21.Tandem_Repeat.tandemRepeatFinder_annotation.region",
"mm9_tandemRepeat",
[],
[],
"max",
None,
True)
dm.deriveNewDataset("mm9.21.Tandem_Repeat.tandemRepeatFinder_annotation_downsampledTo500k.region",
"mm9_tandemRepeat",
[],
[],
"max",
None,
True,
True,
500000)
def mm9_27():
dm.regionDatasets.update(regionDatasetsRepeatsMM9)
dm.deriveNewDataset("mm9.27.Repeat_SINE.repeatMasker_annotation_short_interspersed_nucleotide_elements.region",
"mm9_Repeat_SINE",
[],
[],
None,
None,
True,
True)
dm.deriveNewDataset("mm9.27.Repeat_SINE.repeatMasker_annotation_short_interspersed_nucleotide_elements_downsampledTo500k.region",
"mm9_Repeat_SINE",
[],
[],
None,
None,
True,
True,
500000)
def mm9_26():
dm.regionDatasets.update(regionDatasetsRepeatsMM9)
dm.deriveNewDataset("mm9.26.Repeat_Simple.repeatMasker_annotation_simple_repeats.region",
"mm9_Repeat_Simple",
[],
[],
None,
None,
True,
True)
dm.deriveNewDataset("mm9.26.Repeat_Simple.repeatMasker_annotation_simple_repeats_downsampledTo500k.region",
"mm9_Repeat_Simple",
[],
[],
None,
None,
True,
True,
500000)
