def __init__(self, profile, taxfile, outfile, group_file=None, expand=1):
self.table = TableParser(profile)
self.otu_tax = TaxParser(taxfile)
self.group = parse_group_file(group_file)
if self.group is not None:
self.groups = list(set(self.group.itervalues()))
else:
self.groups = None
self.outfile = outfile
self.expand = expand
self.tax_level = ['kingdom', 'phylum', 'class', 'order', 'family', 'genus', 'species']
self.tax_dict = {
'k': 'kingdom',
'p': 'phylum',
'c': 'class',
'o': 'order',
'f': 'family',
'g': 'genus',
's': 'species',
}
self.samples = {}
self.node_list = {}
self.root_nodes = {}
from __future__ import division
import sys
import os
import argparse
from util import mkdir
this_script_path = os.path.dirname(__file__)
sys.path.insert(1, this_script_path + '/../src')
from Parser import TableParser
def read_params(args):
parser = argparse.ArgumentParser(description='get otu profile uniform | v1.0 at 2015/09/29 by liangzb')
parser.add_argument('-i', '--infile', dest='infile', metavar='FILE', type=str, required=True,
help="set the un-uniformed profile produced by biom convert")
parser.add_argument('-o', '--out_file', dest='outfile', metavar='FILE', type=str, required=True,
help="set the output uniform profile")
args = parser.parse_args()
params = vars(args)
return params
if __name__ == '__main__':
params = read_params(sys.argv)
mkdir(os.path.dirname(params['outfile']))
table = TableParser(params['infile'])
table.parse_table()
table.get_uniform()
table.write_table(params['outfile'])
class Subject(object):
def __init__(self, profile, taxfile, outfile, group_file=None, expand=1):
self.table = TableParser(profile)
self.otu_tax = TaxParser(taxfile)
self.group = parse_group_file(group_file)
if self.group is not None:
self.groups = list(set(self.group.itervalues()))
else:
self.groups = None
self.outfile = outfile
self.expand = expand
self.tax_level = ['kingdom', 'phylum', 'class', 'order', 'family', 'genus', 'species']
self.tax_dict = {
'k': 'kingdom',
'p': 'phylum',
'c': 'class',
'o': 'order',
'f': 'family',
'g': 'genus',
's': 'species',
}
self.samples = {}
self.node_list = {}
self.root_nodes = {}
def read_table(self):
self.table.parse_table()
self.table.get_uniform()
self.samples = self.table.sample_list
def read_tax(self):
self.otu_tax.parse()
self.otu_tax.remove_other()
self.otu_tax = self.otu_tax.OTUs
def get_tax_tree(self):
for otu, otu_tax in self.otu_tax.iteritems():
root_level = self.tax_level[0]
father_level = root_level
try:
tax_name = otu_tax[root_level]
except KeyError:
if otu_tax:
print otu_tax
raise
else:
continue
if tax_name not in self.root_nodes:
root_node = Node(root_level, tax_name, self.samples, groups=self.groups)
self.node_list[tax_name] = root_node
self.root_nodes[tax_name] = root_node
levels = list(otu_tax.iterkeys())
for ind, level in enumerate(levels):
if level not in otu_tax:
continue
father_node = self.node_list[otu_tax[father_level]]
father_level = level
tax_name = otu_tax[level]
if tax_name in self.node_list:
continue
node = Node(level, tax_name, self.samples, groups=self.groups)
self.node_list[tax_name] = node
if tax_name not in father_node.sub_node:
father_node.sub_node[tax_name] = node
# TODO:
# simplify beblow functions
#
def get_node_profile(self, node):
for sample in self.samples:
for otu, otu_tax in self.otu_tax.iteritems():
if node.level not in otu_tax or otu_tax[node.level] != node.name:
continue
try:
node.profile[sample] += self.table.samples[sample][otu]
except KeyError:
continue
if not node.sub_node:
return None
for sub_node in node.sub_node.itervalues():
self.get_node_profile(sub_node)
def get_profile(self):
for node in self.root_nodes.itervalues():
self.get_node_profile(node)
def get_node_group_profile(self, node):
deleted = set()
for sample, group in self.group.iteritems():
for otu, otu_tax in self.otu_tax.iteritems():
if node.level not in otu_tax or otu_tax[node.level] != node.name:
continue
try:
node.group_profile[group] += self.table.samples[sample][otu]
except KeyError, ex:
sample_name = str(ex).strip("'")
if sample_name not in deleted:
sys.stderr.write(sample_name)
deleted.add(sample_name)
continue
for d in deleted:
del self.group[d]
if not node.sub_node:
return None
for sub_node in node.sub_node.itervalues():
self.get_node_group_profile(sub_node)
parser = argparse.ArgumentParser(
description='get otu profile uniform | v1.0 at 2015/09/29 by liangzb')
parser.add_argument(
'-i',
'--infile',
dest='infile',
metavar='FILE',
type=str,
required=True,
help="set the un-uniformed profile produced by biom convert")
parser.add_argument('-o',
'--out_file',
dest='outfile',
metavar='FILE',
type=str,
required=True,
help="set the output uniform profile")
args = parser.parse_args()
params = vars(args)
return params
if __name__ == '__main__':
params = read_params(sys.argv)
mkdir(os.path.dirname(params['outfile']))
table = TableParser(params['infile'])
table.parse_table()
table.get_uniform()
table.write_table(params['outfile'])