8 import sqlite3 as sqlite
15 print "buildrmaskdb: version 2.1"
17 print "usage: python %s rmaskdir rmaskdbfile" % argv[0]
23 buildrmaskdb(rmaskdir, rmaskdb)
26 def buildrmaskdb(rmaskdir, rmaskdb):
27 files = os.listdir(rmaskdir)
28 db = sqlite.connect(rmaskdb)
30 sql.execute("create table repeats (chrom varchar, start int, stop int, name varchar, family varchar)")
31 sql.execute("PRAGMA temp_store = MEMORY")
32 sql.execute("PRAGMA DEFAULT_CACHE_SIZE = 500000")
35 for filename in files:
36 if "rmsk" not in filename:
40 infile = open(rmaskdir + "/" + filename)
42 fields = entry.strip().split("\t")
44 start = int(fields[6])
48 stmt = "insert into repeats values('%s', %d, %d, '%s', '%s')" % (chrom, start, stop, name, family)
53 print "building index..."
54 sql.execute("PRAGMA SYNCHRONOUS = OFF")
55 sql.execute("create index chromIndex on repeats(chrom)")
56 sql.execute("create index mainIndex on repeats(chrom, start, stop)")
60 if __name__ == "__main__":