def _trackhub_url_info(self, j):
assembly = self.assembly = j["assembly"]
pgSearch = PGsearch(self.pw, assembly)
if "coord_start" not in j:
cre = CRE(pgSearch, j["accession"], self.cacheW[assembly])
coord = cre.coord()
else:
coord = Coord(j["coord_chrom"], j["coord_start"], j["coord_end"])
coord.resize(j["halfWindow"])
return assembly, j["accession"], coord
def parseSnp(self):
snps = self.dbsnps.lookup(self.assembly, self.loci)
if not snps:
return None
if len(snps) > 1:
# search on UCSC
return None
snp = snps[0]
c = Coord(snp[0], snp[1], snp[2])
c.resize(self.halfWindow)
return c
def parseSnp(self):
snps = self.dbsnps.lookup(self.assembly, self.loci)
if not snps:
return None
if len(snps) > 1:
# search on UCSC
return None
snp = snps[0]
c = Coord(snp[0], snp[1], snp[2])
c.resize(self.halfWindow)
return c
def ucsc_trackhub_url_snp(self, j, uuid):
assembly = self.assembly = j["assembly"]
pgSearch = PGsearch(self.pw, assembly)
snp = j["snp"]
c = Coord(snp["chrom"], snp["cre_start"], snp["cre_end"])
hubNum = self.db.insertOrUpdate(assembly, snp["accession"], uuid, j)
trackhubUrl = '/'.join([j["host"],
"ucsc_trackhub",
uuid,
"hub_" + str(hubNum) + ".txt"])
url = "https://genome.ucsc.edu/cgi-bin/hgTracks?"
url += "db=" + assembly
url += "&position=" + str(c)
url += "&hubClear=" + trackhubUrl
url += "&highlight=" + assembly + "." + c.chrom + "%3A" + str(snp["snp_start"]) + '-' + str(snp["snp_end"])
if "hg19" == assembly:
url += "&g=wgEncodeGencodeV19"
url += "&g=phastCons100way"
return {"url": url, "trackhubUrl": trackhubUrl}
def ucsc_trackhub_url(self, j, uuid):
assembly, accession, coord = self._trackhub_url_info(j)
hubNum = self.db.insertOrUpdate(assembly, accession, uuid, j)
c = Coord(j["coord_chrom"], j["coord_start"], j["coord_end"])
trackhubUrl = '/'.join([j["host"],
"ucsc_trackhub",
uuid,
"hub_" + str(hubNum) + ".txt"])
url = "https://genome.ucsc.edu/cgi-bin/hgTracks?"
url += "db=" + assembly
url += "&position=" + str(coord)
url += "&hubClear=" + trackhubUrl
url += "&highlight=" + assembly + "." + c.chrom + "%3A" + str(c.start) + '-' + str(c.end)
if "hg19" == assembly:
url += "&g=wgEncodeGencodeV19"
url += "&g=phastCons100way"
if "mm10" == assembly:
# FIXME
pass
return {"url": url, "trackhubUrl": trackhubUrl}
def parseGene(self):
genes = self.genes.lookup(self.assembly, self.loci)
if not genes:
genes = self.genes.fuzzy_lookup(self.assembly, self.loci)
if not genes:
self.userErrMsg = "'{loci}' not found".format(loci=self.loci)
else:
self.userErrMsg = "'{loci}' not found; potential matches: {genes}".format(loci=self.loci, genes=", ".join(sorted(genes)))
return None
if len(genes) > 1:
self.userErrMsg = "Multiple genomic positions found; using first found..."
return None
gene = genes[0]
c = Coord(gene[0], gene[1], gene[2])
c.resize(self.halfWindow)
return c
def parseGene(self):
genes = self.genes.lookup(self.assembly, self.loci)
if not genes:
genes = self.genes.fuzzy_lookup(self.assembly, self.loci)
if not genes:
self.userErrMsg = "'{loci}' not found".format(loci=self.loci)
else:
self.userErrMsg = "'{loci}' not found; potential matches: {genes}".format(
loci=self.loci, genes=", ".join(sorted(genes)))
return None
if len(genes) > 1:
self.userErrMsg = "Multiple genomic positions found; using first found..."
return None
gene = genes[0]
c = Coord(gene[0], gene[1], gene[2])
c.resize(self.halfWindow)
return c
def getRankedPeakCoord(self):
if 1 != len(self.tissue_ids):
self.userErrMsg = "Please only select one tissue"
return None
wepis = self.epigenomes.GetByAssemblyAndAssays(self.assembly,
self.assays)
wepis = filter(lambda e: e.web_id() in self.tissue_ids, wepis.epis)
if 1 != len(wepis):
self.userErrMsg = "Please only select one tissue"
return None
wepi = wepis[0]
fnp = wepi.predictionFnp().replace(".bigBed", ".bed")
if not os.path.exists(fnp):
raise Exception("file not found " + fnp)
rank = int(self.loci) - 1
if rank < 0:
self.userErrMsg = "Ranks start at 0"
return None
row = None
with open(fnp) as f:
# http://stackoverflow.com/a/2081880
for i, line in enumerate(f):
if i == rank:
row = line
break
if i > rank:
break
if not row:
self.userErrMsg = "Only %d peaks in file" % i
return None
toks = row.rstrip().split('\t')
c = Coord(toks[0], toks[1], toks[2])
c.resize(self.halfWindow)
return c
def getRankedPeakCoord(self):
if 1 != len(self.tissue_ids):
self.userErrMsg = "Please only select one tissue"
return None
wepis = self.epigenomes.GetByAssemblyAndAssays(self.assembly, self.assays)
wepis = filter(lambda e: e.web_id() in self.tissue_ids, wepis.epis)
if 1 != len(wepis):
self.userErrMsg = "Please only select one tissue"
return None
wepi = wepis[0]
fnp = wepi.predictionFnp().replace(".bigBed", ".bed")
if not os.path.exists(fnp):
raise Exception("file not found " + fnp)
rank = int(self.loci) - 1
if rank < 0:
self.userErrMsg = "Ranks start at 0"
return None
row = None
with open(fnp) as f:
# http://stackoverflow.com/a/2081880
for i, line in enumerate(f):
if i == rank:
row = line
break
if i > rank:
break
if not row:
self.userErrMsg = "Only %d peaks in file" % i
return None
toks = row.rstrip().split('\t')
c = Coord(toks[0], toks[1], toks[2])
c.resize(self.halfWindow)
return c
def search(self):
coord = None
if self.loci.lower().startswith("chr"):
# coordinate
coord = Coord.parse(self.loci)
elif self.loci.lower().startswith("rs"):
# SNP
coord = self.parseSnp()
elif self.loci.isdigit():
# a ranked peak for a single selected tissue
coord = self.getRankedPeakCoord()
else:
coord = self.parseGene()
return coord
def search(self):
coord = None
if self.loci.lower().startswith("chr"):
# coordinate
coord = Coord.parse(self.loci)
elif self.loci.lower().startswith("rs"):
# SNP
coord = self.parseSnp()
elif self.loci.isdigit():
# a ranked peak for a single selected tissue
coord = self.getRankedPeakCoord()
else:
coord = self.parseGene()
return coord