diff options
Diffstat (limited to 'scripts')
| -rwxr-xr-x | scripts/addRif.py | 131 |
1 files changed, 67 insertions, 64 deletions
diff --git a/scripts/addRif.py b/scripts/addRif.py index 8ea5f74..4d33af3 100755 --- a/scripts/addRif.py +++ b/scripts/addRif.py @@ -26,70 +26,69 @@ # created by Lei Yan 02/08/2011 -import string -import MySQLdb -import time import os import sys +import MySQLdb path1 = os.path.abspath(os.path.dirname(__file__)) path2 = path1 + "/.." path3 = path1 + "/../../tmp" sys.path.insert(0, path2) + def fetchrif(): - try: - con = MySQLdb.Connect(db="gn3", host="localhost", user="gn2", passwd="password") - cursor = con.cursor() - print("You have successfully connected to mysql.\n") - except: - print("You entered incorrect password.\n") - sys.exit(0) + try: + con = MySQLdb.Connect(db="gn3", host="localhost", user="gn2", passwd="password") + cursor = con.cursor() + print("You have successfully connected to mysql.\n") + except: + print("You entered incorrect password.\n") + sys.exit(0) - taxIds = {'10090':1, '9606':4, '10116':2, '3702':3} - taxIdKeys = taxIds.keys() + taxIds = {"10090": 1, "9606": 4, "10116": 2, "3702": 3} + taxIdKeys = taxIds.keys() - os.chdir(path3) - print("path3: %s" % (path3)) - genedict = {} + os.chdir(path3) + print("path3: %s" % (path3)) + genedict = {} - os.system("rm -vf gene_info") - os.system("wget ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_info.gz") - os.system("gunzip gene_info.gz") + os.system("rm -vf gene_info") + os.system("wget ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_info.gz") + os.system("gunzip gene_info.gz") - file = open("gene_info", 'r') - i = 0 - for line1 in file: - line1 = line1.strip() - if line1.startswith('#'): - continue - line2 = line1.strip().split("\t") - if line2[0] in taxIdKeys: - genedict[line2[1]] = line2[2] - i += 1 - if i%1000000 == 0: - print("finished: %d" % (i)) - print("finished all: %d" % (i)) - file.close() + file = open("gene_info", "r") + i = 0 + for line1 in file: + line1 = line1.strip() + if line1.startswith("#"): + continue + line2 = line1.strip().split("\t") + if line2[0] in taxIdKeys: + genedict[line2[1]] = line2[2] + i += 1 + if i % 1000000 == 0: + print("finished: %d" % (i)) + print("finished all: %d" % (i)) + file.close() - os.system("rm -vf generifs_basic") - os.system("wget ftp://ftp.ncbi.nlm.nih.gov/gene/GeneRIF/generifs_basic.gz") - os.system("gunzip generifs_basic.gz") + os.system("rm -vf generifs_basic") + os.system("wget ftp://ftp.ncbi.nlm.nih.gov/gene/GeneRIF/generifs_basic.gz") + os.system("gunzip generifs_basic.gz") - file = open("generifs_basic", 'r') - i = 0 - for line1 in file: - line1 = line1.strip() - if line1.startswith('#'): - continue - line2 = line1.strip().split("\t") - if line2[0] in taxIdKeys and len(line2) >= 5: - line2[0] = taxIds[line2[0]] - try: - symbol = genedict[line2[1]] - except: - symbol = "" - sql = """ + file = open("generifs_basic", "r") + i = 0 + for line1 in file: + line1 = line1.strip() + if line1.startswith("#"): + continue + line2 = line1.strip().split("\t") + if line2[0] in taxIdKeys and len(line2) >= 5: + line2[0] = taxIds[line2[0]] + try: + symbol = genedict[line2[1]] + except: + symbol = "" + sql = """ SELECT COUNT(*) FROM GeneRIF_BASIC WHERE GeneRIF_BASIC.`SpeciesId`=%s @@ -98,11 +97,11 @@ def fetchrif(): AND GeneRIF_BASIC.`createtime`=%s AND GeneRIF_BASIC.`comment`=%s """ - cursor.execute(sql, (line2[0], line2[1], line2[2], line2[3], line2[4])) - c = cursor.fetchone()[0] - if c == 0: - print("to insert...") - sql = """ + cursor.execute(sql, (line2[0], line2[1], line2[2], line2[3], line2[4])) + c = cursor.fetchone()[0] + if c == 0: + print("to insert...") + sql = """ INSERT INTO GeneRIF_BASIC SET GeneRIF_BASIC.`SpeciesId`=%s, GeneRIF_BASIC.`GeneId`=%s, @@ -111,16 +110,20 @@ def fetchrif(): GeneRIF_BASIC.`createtime`=%s, GeneRIF_BASIC.`comment`=%s """ - cursor.execute(sql, (line2[0], line2[1], symbol, line2[2], line2[3], line2[4])) - i += 1 - if i%100000 == 0: - print("finished: %d" % (i)) - print("finished all: %d" % (i)) - file.close() - cursor.close() + cursor.execute( + sql, (line2[0], line2[1], symbol, line2[2], line2[3], line2[4]) + ) + i += 1 + if i % 100000 == 0: + print("finished: %d" % (i)) + print("finished all: %d" % (i)) + file.close() + cursor.close() + + # /usr/bin/python addRif.py if __name__ == "__main__": - print("command line arguments:\n\t%s" % sys.argv) - fetchrif() - print("exit successfully") + print("command line arguments:\n\t%s" % sys.argv) + fetchrif() + print("exit successfully") |