class StaphAureusN315Recipe(BaseRecipe):
def __init__(self):
super(StaphAureusN315Recipe, self).__init__()
self.source = SourceFile(self.repo, "staph_aureus_n315.fna.gz",
"https://www.ncbi.nlm.nih.gov/nuccore/NC_002745.2?report=fasta&log$=seqview&format=text")
self.bt2 = ConstructedFile(self.repo, "staph_aureus_n315.bt2")
def name(self):
return 'staph_aureus_n315'
def fileTypes(self):
return ['gz_fasta_nucl', 'bt2_index']
def resultSchema(self):
return {
'fasta': 'gz_fasta_nucl',
'bt2': ['bt2_index'] * 6
}
def makeRecipe(self):
self.source.resolve()
self.repo.saveFiles(self,
'fasta',
self.source.filepath())
self.bt2.resolve()
bt2Indices = glob(self.bt2.filepath() + '*')
self.repo.saveFiles(self,
'bt2',
*bt2Indices)
class CARDRecipe(BaseRecipe):
'''
Recipe for the comprehensive antibiotic resistance
database
'''
def __init__(self):
super(CARDRecipe, self).__init__()
self.source = SourceFile(self.repo, "card.faa.gz")
self.sbred = ConstructedFile(self.repo, "card.shortbred_markers.faa")
self.dmnd = ConstructedFile(self.repo, "card.dmnd")
def name(self):
return 'card'
def fileTypes(self):
return ['gz_fasta_aa', 'fasta_aa', 'dmnd-db']
def resultSchema(self):
return {
'fasta': 'gz_fasta_aa',
'sbred': 'fasta_aa',
'dmnd': 'dmnd-db',
}
def makeRecipe(self):
self.source.resolve()
self.repo.saveFiles(self, 'fasta', self.source.filepath())
self.sbred.resolve()
self.repo.saveFiles(self, 'sbred', self.sbred.filepath())
self.dmnd.resolve()
self.repo.saveFiles(self, 'dmnd', self.dmnd.filepath())
class MiniKrakenRecipe(BaseRecipe):
'''
Recipe for minikraken db
'''
def __init__(self):
super(MiniKrakenRecipe, self).__init__()
self.source = SourceFile(self.repo, "minikraken-db")
def name(self):
return 'minikraken'
def fileTypes(self):
return ['kraken-db']
def resultSchema(self):
return {'kraken-db': ['kraken-db'] * 4}
def makeRecipe(self):
self.source.resolve()
dbIdx = os.path.join(self.source.filepath(), 'database.idx')
kDb = os.path.join(self.source.filepath(), 'database.kdb')
taxa = os.path.join(self.source.filepath(), 'taxonomy')
taxaNames = os.path.join(taxa, 'names.dmp')
taxaNodes = os.path.join(taxa, 'nodes.dmp')
fs = [dbIdx, kDb, taxaNames, taxaNodes]
self.repo.saveFiles(self, 'kraken-db', *fs)
def __init__(self):
super(GOTTCHARecipe, self).__init__()
self.bact_species = SourceFile(
self.repo, "GOTTCHA_BACTERIA_c4937_k24_u30.species")
self.bact_strains = SourceFile(
self.repo, "GOTTCHA_BACTERIA_c4937_k24_u30.strain")
self.virus_species = SourceFile(
self.repo, "GOTTCHA_VIRUSES_c5900_k24_u30.species")
self.virus_strains = SourceFile(
self.repo, "GOTTCHA_VIRUSES_c5900_k24_u30.strain")
class HG38UCSCGenomeRecipe(BaseRecipe):
'''
Recipe for the hg38 genome from UCSC.
This genome is used in the CAP because:
- hg38 is the most recent as of December 2017
- it includes alt contigs. Bad for variant calling, good for filtering
- it uses human readable chromosome names
- it uses the rCRS mitochondrial sequence
'''
def __init__(self):
super(HG38UCSCGenomeRecipe, self).__init__()
self.source = SourceFile(self.repo, "hg38_ucsc.fna.gz",
"http://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/hg38.fa.gz")
self.bt2 = ConstructedFile(self.repo, "hg38_ucsc.bt2")
def name(self):
return 'hg38_ucsc'
def fileTypes(self):
return ['gz_fasta_nucl', 'bt2_index']
def resultSchema(self):
return {
'fasta': 'gz_fasta_nucl',
'bt2': ['bt2_index'] * 6
}
def makeRecipe(self):
self.source.resolve()
self.repo.saveFiles(self,
'fasta',
self.source.filepath())
self.bt2.resolve()
bt2Indices = glob(self.bt2.filepath() + '*')
self.repo.saveFiles(self,
'bt2',
*bt2Indices)
class NRProteinRecipe(BaseRecipe):
'''
Recipe for uniref90 with diamond index
'''
def __init__(self):
super(NRProteinRecipe, self).__init__()
self.fasta = SourceFile(self.repo, "nr.faa.gz")
self.dmnd = ConstructedFile(self.repo, "nr.dmnd")
def name(self):
return 'nr_protein'
def fileTypes(self):
return ['gz_fasta_aa', 'dmnd-db']
def resultSchema(self):
return {'fasta': 'gz_fasta_aa', 'dmnd': 'dmnd-db'}
def makeRecipe(self):
self.fasta.resolve()
self.repo.saveFiles(self, 'fasta', self.fasta.filepath())
self.dmnd.resolve()
self.repo.saveFiles(self, 'dmnd', self.dmnd.filepath())
class MeganRecipe(BaseRecipe):
'''
Recipe for uniref90 with diamond index
'''
def __init__(self):
super(MeganRecipe, self).__init__()
self.acc2taxa = SourceFile(self.repo, "prot_acc2tax-May2017.abin")
self.blast2lca = ConstructedFile(self.repo, "blast2lca")
def name(self):
return 'megan'
def fileTypes(self):
return ['megan_table']
def resultSchema(self):
return {'acc2taxa': 'megan_table', 'blast2lca': 'megan_table'}
def makeRecipe(self):
self.acc2taxa.resolve()
self.repo.saveFiles(self, 'acc2taxa', self.acc2taxa.filepath())
self.blast2lca.resolve()
self.repo.saveFiles(self, 'blast2lca', self.blast2lca.filepath())
class CommonMacrobialRecipe(BaseRecipe):
'''
Recipe for database of common macrobial genomes
'''
def __init__(self):
super(CommonMacrobialRecipe, self).__init__()
self.source = SourceFile(self.repo, "macrobes.fna.gz")
self.bt2 = ConstructedFile(self.repo, "common_macrobial.bt2")
def name(self):
return 'common_macrobial'
def fileTypes(self):
return ['gz_fasta_nucl', 'bt2_index']
def resultSchema(self):
return {'fasta': 'gz_fasta_nucl', 'bt2': ['bt2_index'] * 6}
def makeRecipe(self):
self.source.resolve()
self.repo.saveFiles(self, 'fasta', self.source.filepath())
self.bt2.resolve()
bt2Indices = glob(self.bt2.filepath() + '*')
self.repo.saveFiles(self, 'bt2', *bt2Indices)
class MethyltransferaseRecipe(BaseRecipe):
'''
Recipe for the comprehensive antibiotic resistance
database
'''
def __init__(self):
super(MethyltransferaseRecipe, self).__init__()
self.fasta = SourceFile(self.repo, "methyls.faa")
self.dmnd = ConstructedFile(self.repo, "methyls.dmnd")
def name(self):
return 'methyl'
def fileTypes(self):
return ['gz_fasta_aa', 'dmnd-db']
def resultSchema(self):
return {'fasta': 'gz_fasta_aa', 'dmnd': 'dmnd-db'}
def makeRecipe(self):
self.fasta.resolve()
self.repo.saveFiles(self, 'fasta', self.fasta.filepath())
self.dmnd.resolve()
self.repo.saveFiles(self, 'dmnd', self.dmnd.filepath())
class MegaresRecipe(BaseRecipe):
def __init__(self):
super(MegaresRecipe, self).__init__()
self.fasta = SourceFile(self.repo, "megares.fa")
self.bt2 = ConstructedFile(self.repo, "megares.bt2")
self.csv = SourceFile(self.repo, "megares.csv")
def name(self):
return 'megares'
def fileTypes(self):
return ['fasta_nucl', 'bt2_index', 'csv']
def resultSchema(self):
return {'fasta': 'fasta_nucl', 'bt2': ['bt2_index'] * 6, 'csv': 'csv'}
def makeRecipe(self):
self.fasta.resolve()
self.repo.saveFiles(self, 'fasta', self.fasta.filepath())
self.csv.resolve()
self.repo.saveFiles(self, 'csv', self.csv.filepath())
self.bt2.resolve()
bt2Indices = glob(self.bt2.filepath() + '*')
self.repo.saveFiles(self, 'bt2', *bt2Indices)
def __init__(self):
super(CentrifugeRecipe, self).__init__()
self.nt = SourceFile(self.repo, "nt.1.cf")
self.phv = SourceFile(self.repo, "p+h+v.1.cf")
def __init__(self):
super(HG38UCSCGenomeRecipe, self).__init__()
self.source = SourceFile(self.repo, "hg38_ucsc.fna.gz",
"http://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/hg38.fa.gz")
self.bt2 = ConstructedFile(self.repo, "hg38_ucsc.bt2")
def __init__(self):
super(VFDBRecipe, self).__init__()
self.fasta = SourceFile(self.repo, "vfdb.faa")
self.dmnd = ConstructedFile(self.repo, "vfdb.dmnd")
def __init__(self):
super(KrakenHLLRecipe, self).__init__()
self.refseq = SourceFile(self.repo, "refseq-db")
self.nt = SourceFile(self.repo, "nt")
def __init__(self):
super(NRProteinRecipe, self).__init__()
self.fasta = SourceFile(self.repo, "nr.faa.gz")
self.dmnd = ConstructedFile(self.repo, "nr.dmnd")
def __init__(self):
super(StaphAureusN315Recipe, self).__init__()
self.source = SourceFile(self.repo, "staph_aureus_n315.fna.gz",
"https://www.ncbi.nlm.nih.gov/nuccore/NC_002745.2?report=fasta&log$=seqview&format=text")
self.bt2 = ConstructedFile(self.repo, "staph_aureus_n315.bt2")
def __init__(self):
super(CommonMacrobialRecipe, self).__init__()
self.source = SourceFile(self.repo, "macrobes.fna.gz")
self.bt2 = ConstructedFile(self.repo, "common_macrobial.bt2")
def __init__(self):
super(MiniKrakenRecipe, self).__init__()
self.source = SourceFile(self.repo, "minikraken-db")
def __init__(self):
super(MegaresRecipe, self).__init__()
self.fasta = SourceFile(self.repo, "megares.fa")
self.bt2 = ConstructedFile(self.repo, "megares.bt2")
self.csv = SourceFile(self.repo, "megares.csv")
def __init__(self):
super(CARDRecipe, self).__init__()
self.source = SourceFile(self.repo, "card.faa.gz")
self.sbred = ConstructedFile(self.repo, "card.shortbred_markers.faa")
self.dmnd = ConstructedFile(self.repo, "card.dmnd")
def __init__(self):
super(MeganRecipe, self).__init__()
self.acc2taxa = SourceFile(self.repo, "prot_acc2tax-May2017.abin")
self.blast2lca = ConstructedFile(self.repo, "blast2lca")
def __init__(self):
super(MethyltransferaseRecipe, self).__init__()
self.fasta = SourceFile(self.repo, "methyls.faa")
self.dmnd = ConstructedFile(self.repo, "methyls.dmnd")