def write(self,blastMap=None, transcriptMin=9, transcriptMax=1000,outFile="genesets.gmt"):
"""
outFile: specifies the output file path (*.gmt)
also a *.csv file with gene transcript mapping will be created if a bmap is provided
blastMap: BlastMap returned after loading summary file in BlastMapper
transcriptMin: minimum size for a gene set
transcriptMax: maximum size for a gene set
outFile: outfile path
"""
print("---------------------")
if self.gene2go:
print('There are %s genes with at least one annotation'%(len(self.gene2go.keys())))
print('There are %s genes in the labels file'%(len(self.genes)))
if blastMap:
bm = BlastMapper()
bmGenes = bm.print_summary(blastMap)
gene2transcript = bm.get_gene_dict(blastMap)
if self.gene2go:
usableGenes = list(set(bmGenes).intersection(set(self.gene2go.keys())))
else:
usableGenes = bmGenes
if blastMap:
print('There are %s genes with at least one BLAST hit'%(len(bmGenes)))
print('There are %s genes that have both a BLAST hit and an annotation'%(len(usableGenes)))
#print('There are %s genes in clusters with at least one BLAST hits'%(len(set(self.genes).intersection.(set(bmGenes.keys())))))
## prepare outfiles
writer = csv.writer(open(outFile,'w'),delimiter="\t")
if blastMap:
outFileMap = re.sub("\.gmt",".csv",outFile)
writerMap = csv.writer(open(outFileMap,'w'))
writerMap.writerow(["gene_set","gene_id","mapped_transcripts"])
## save gene sets to file
failedThreshold = 0
for _k in self.allClusters:
clusterInds = np.where(self.labels==_k)[0]
clusterGenes = self.genes[clusterInds]
gsName = "gs-"+str(_k)
if self.gene2go:
description = self.get_description(clusterGenes)
else:
description = "kegg pathway"
## map the genes
if blastMap:
mapped = set([])
for gene in clusterGenes:
if not gene2transcript.has_key(gene):
continue
geneTranscripts = gene2transcript[gene]
geneTranscripts = list(set([re.sub("\.[0-9]$","",g) for g in geneTranscripts]))
if blastMap:
writerMap.writerow([gsName,gene,";".join(list(geneTranscripts))])
mapped.update(geneTranscripts)
mapped = list(mapped)
else:
mapped = clusterGenes
### remove non-unique and versioned genes
#if len(mapped) > 0:
if len(mapped) >= transcriptMin and len(mapped) <= transcriptMax:
writer.writerow([gsName,description] + mapped)
else:
failedThreshold+=clusterGenes.size
print("-----------------")
#print("sigma: %s"%self.sigma)
#print("k: %s"%self.k)
print('Total clusters: %s '%self.allClusters.size)
percentAccepted = float(self.genes.size-failedThreshold) / float(self.genes.size)
print("genes pass threshold %s/%s (%s)"%(self.genes.size-failedThreshold,self.genes.size,round(percentAccepted,2)) + "%)")
def write(self,
blastMap=None,
transcriptMin=9,
transcriptMax=1000,
outFile="genesets.gmt"):
"""
outFile: specifies the output file path (*.gmt)
also a *.csv file with gene transcript mapping will be created if a bmap is provided
blastMap: BlastMap returned after loading summary file in BlastMapper
transcriptMin: minimum size for a gene set
transcriptMax: maximum size for a gene set
outFile: outfile path
"""
print("---------------------")
if self.gene2go:
print('There are %s genes with at least one annotation' %
(len(self.gene2go.keys())))
print('There are %s genes in the labels file' % (len(self.genes)))
if blastMap:
bm = BlastMapper()
bmGenes = bm.print_summary(blastMap)
gene2transcript = bm.get_gene_dict(blastMap)
if self.gene2go:
usableGenes = list(
set(bmGenes).intersection(set(self.gene2go.keys())))
else:
usableGenes = bmGenes
if blastMap:
print('There are %s genes with at least one BLAST hit' %
(len(bmGenes)))
print(
'There are %s genes that have both a BLAST hit and an annotation'
% (len(usableGenes)))
#print('There are %s genes in clusters with at least one BLAST hits'%(len(set(self.genes).intersection.(set(bmGenes.keys())))))
## prepare outfiles
writer = csv.writer(open(outFile, 'w'), delimiter="\t")
if blastMap:
outFileMap = re.sub("\.gmt", ".csv", outFile)
writerMap = csv.writer(open(outFileMap, 'w'))
writerMap.writerow(["gene_set", "gene_id", "mapped_transcripts"])
## save gene sets to file
failedThreshold = 0
for _k in self.allClusters:
clusterInds = np.where(self.labels == _k)[0]
clusterGenes = self.genes[clusterInds]
gsName = "gs-" + str(_k)
if self.gene2go:
description = self.get_description(clusterGenes)
else:
description = "kegg pathway"
## map the genes
if blastMap:
mapped = set([])
for gene in clusterGenes:
if not gene2transcript.has_key(gene):
continue
geneTranscripts = gene2transcript[gene]
geneTranscripts = list(
set([
re.sub("\.[0-9]$", "", g) for g in geneTranscripts
]))
if blastMap:
writerMap.writerow(
[gsName, gene, ";".join(list(geneTranscripts))])
mapped.update(geneTranscripts)
mapped = list(mapped)
else:
mapped = clusterGenes
### remove non-unique and versioned genes
if type(mapped) == type(np.array([])):
mapped = mapped.tolist()
if len(mapped) >= transcriptMin and len(mapped) <= transcriptMax:
writer.writerow([gsName, description] + mapped)
else:
failedThreshold += clusterGenes.size
print("-----------------")
#print("sigma: %s"%self.sigma)
#print("k: %s"%self.k)
print('Total clusters: %s ' % self.allClusters.size)
percentAccepted = float(self.genes.size - failedThreshold) / float(
self.genes.size)
print(
"genes pass threshold %s/%s (%s)" %
(self.genes.size - failedThreshold, self.genes.size,
round(percentAccepted, 2)) + "%)")
