def buildGeneListMatrix(infiles, outfile):
'''build a gene list matrix for simple pathway analysis
based on hypergeometric test.
A gene list is derived from a gene set by
applying thresholds to the input data set. The
thresholds are defined in the configuration file.
'''
genesets = []
backgrounds = []
headers = []
for infile in infiles:
genelist = pandas.read_csv(IOTools.openFile(infile),
index_col=0,
sep='\t')
track = P.snip(os.path.basename(infile), ".tsv.gz")
headers.append(track)
field = PARAMS[P.matchParameter("%s_foreground_field" % track)]
min_threshold = PARAMS[P.matchParameter("%s_foreground_min_threshold" %
track)]
max_threshold = PARAMS[P.matchParameter("%s_foreground_max_threshold" %
track)]
genesets.append(
set(genelist[(genelist[field] >= min_threshold)
& (genelist[field] <= max_threshold)].index))
E.info('%s: foreground: %f <= %s <= %f' %
(track, min_threshold, field, max_threshold))
field = PARAMS[P.matchParameter("%s_background_field" % track)]
min_threshold = PARAMS[P.matchParameter("%s_background_min_threshold" %
track)]
max_threshold = PARAMS[P.matchParameter("%s_background_max_threshold" %
track)]
E.info('%s: background: %f <= %s <= %f' %
(track, min_threshold, field, max_threshold))
backgrounds.append(
set(genelist[(genelist[field] >= min_threshold)
& (genelist[field] <= max_threshold)].index))
E.info("%s: fg=%i, bg=%i" %
(track, len(genesets[-1]), len(backgrounds[-1])))
E.info("writing gene list matrix")
with IOTools.openFile(outfile, "w") as outf:
SetTools.writeSets(outf, genesets, labels=headers)
with IOTools.openFile(outfile + ".bg.tsv.gz", "w") as outf:
SetTools.writeSets(outf, backgrounds, labels=headers)
E.info("writing intersection/union matrix")
# build set intersection matrix
matrix = SetTools.unionIntersectionMatrix(genesets)
with IOTools.openFile(outfile + ".matrix.gz", "w") as outf:
IOTools.writeMatrix(outf, matrix, headers, headers)
matrix = SetTools.unionIntersectionMatrix(backgrounds)
with IOTools.openFile(outfile + ".bg.matrix.gz", "w") as outf:
IOTools.writeMatrix(outf, matrix, headers, headers)
