def read_header(self):
if self.header == None:
line = FileReader.readline(self).rstrip()
hdr = line.split("\t")
self.header = MatrixHeader(hdr)
return self.header
def merge(log, input, output, gitools_output):
"""
Merge repeated rows by the lowest pvalue, in case the pvalue is the same take the one with greater n
"""
f = FileReader(input)
hdr = f.readline().rstrip().split("\t")
upreg = {}
downreg = {}
upreg_count = 0
downreg_count = 0
mid_index = 8
for line in f:
line = line.rstrip()
if len(line) == 0:
continue
fields = line.split("\t")
row_name = fields[0]
upreg_count += merge_data(row_name, fields[1:mid_index], upreg)
downreg_count += merge_data(row_name, fields[mid_index:], downreg)
f.close()
upreg_keys = upreg.keys()
downreg_keys = downreg.keys()
log.debug("Total rows: upreg = {}, downreg = {}".format(len(upreg_keys), len(downreg_keys)))
log.debug("Merged rows: upreg = {}, downreg = {}".format(upreg_count, downreg_count))
ofile = FileWriter(output)
ofile.write("\t".join(hdr))
ofile.write("\n")
gfile = FileWriter(gitools_output)
gfile.write("column\trow\t")
gfile.write("\t".join([x[6:] for x in hdr if x.startswith("upreg_")]))
gfile.write("\n")
for row_name in upreg_keys:
upreg_data = upreg[row_name]
upreg_data_join = "\t".join(upreg_data)
downreg_data = downreg[row_name]
downreg_data_join = "\t".join(downreg_data)
ofile.write(row_name)
ofile.write("\t")
ofile.write(upreg_data_join)
ofile.write("\t")
ofile.write(downreg_data_join)
ofile.write("\n")
gfile.write("upreg\t")
gfile.write(row_name)
gfile.write("\t")
gfile.write(upreg_data_join)
gfile.write("\n")
gfile.write("downreg\t")
gfile.write(row_name)
gfile.write("\t")
gfile.write(downreg_data_join)
gfile.write("\n")
ofile.close()
gfile.close()
return (upreg_count, downreg_count)
