mirror of
https://github.com/galaxyproject/galaxy.git
synced 2026-10-25 10:24:38 +08:00
Nicola Soranzo
ee721c5d36
Do not create file in case of errors, so the cron script will not overwrite a good file with a bad one.
80 lines
2.5 KiB
Python
80 lines
2.5 KiB
Python
#!/usr/bin/env python
|
|
|
|
"""
|
|
Connects to a UCSC table browser and scrapes chrominfo for every build
|
|
specified by an input file (such as one output by parse_builds.py).
|
|
If not input file specified, it will connect using parse_builds.py to
|
|
retrieve a list of available builds.
|
|
|
|
All chromInfo is placed in a path with the convention
|
|
{dbpath}/buildname.len
|
|
|
|
Usage:
|
|
python build_chrom_db.py dbpath/ [builds_file]
|
|
"""
|
|
|
|
import fileinput
|
|
import os
|
|
import sys
|
|
import urllib
|
|
|
|
import parse_builds
|
|
|
|
|
|
def getchrominfo(url, db):
|
|
tableURL = "http://genome-test.cse.ucsc.edu/cgi-bin/hgTables?"
|
|
URL = tableURL + urllib.urlencode({
|
|
"clade" : "",
|
|
"org" : "",
|
|
"db" : db,
|
|
"hgta_outputType": "primaryTable",
|
|
"hgta_group" : "allTables",
|
|
"hgta_table" : "chromInfo",
|
|
"hgta_track" : db,
|
|
"hgta_regionType": "",
|
|
"position": "",
|
|
"hgta_doTopSubmit": "get info"})
|
|
page = urllib.urlopen(URL)
|
|
for line in page:
|
|
line = line.rstrip( "\r\n" )
|
|
if line.startswith("#"):
|
|
continue
|
|
fields = line.split("\t")
|
|
if len(fields) > 1 and len(fields[0]) > 0 and int(fields[1]) > 0:
|
|
yield [fields[0], fields[1]]
|
|
else:
|
|
raise Exception("Problem parsing line '%s'" % line)
|
|
|
|
if __name__ == "__main__":
|
|
if len(sys.argv) == 1:
|
|
sys.exit("Path to place chromInfo tables must be specified.")
|
|
dbpath = sys.argv[1]
|
|
builds = []
|
|
if len(sys.argv) > 2:
|
|
try:
|
|
buildfile = fileinput.FileInput(sys.argv[2])
|
|
for line in buildfile:
|
|
if line.startswith("#"):
|
|
continue
|
|
builds.append(line.split("\t")[0])
|
|
except:
|
|
sys.exit("Bad input file.")
|
|
else:
|
|
try:
|
|
for build in parse_builds.getbuilds("http://genome.cse.ucsc.edu/cgi-bin/das/dsn"):
|
|
builds.append(build[0])
|
|
except:
|
|
sys.exit("Unable to retrieve builds.")
|
|
for build in builds:
|
|
if build == "?":
|
|
continue # no lengths for unspecified chrom
|
|
print "Retrieving " + build
|
|
outfile_name = dbpath + build + ".len"
|
|
try:
|
|
with open(outfile_name, "w") as outfile:
|
|
for chrominfo in getchrominfo("http://genome-test.cse.ucsc.edu/cgi-bin/hgTables?", build):
|
|
print >> outfile, "\t".join(chrominfo)
|
|
except Exception as e:
|
|
print "Failed to retrieve %s: %s" % (build, e)
|
|
os.remove(outfile_name)
|