Merge pull request #2 from bbglab/master

use custom config path and use memory map files to speed up genomereference retrival

transvar/config.py

@@ -55,7 +55,7 @@ def gunzip(fn):
 
 
 cfg_fns = [os.path.join(os.path.dirname(__file__), 'transvar.cfg'),
-           os.path.expanduser('~/.transvar.cfg')]
+           os.path.expanduser(os.getenv('TRANSVAR_CFG', '~/.transvar.cfg'))]
 
 downloaddirs = [os.path.join(os.path.dirname(__file__), 'transvar.download'),
                 os.path.expanduser('~/.transvar.download')]

transvar/faidx.py

@@ -1,6 +1,8 @@
 """ faidx python code adapted from Allen Yu 
 http://www.allenyu.info/item/24-quickly-fetch-sequence-from-samtools-faidx-indexed-fasta-sequences.html """
 import sys
+import mmap
+
 from err import *
 from utils import *
 
@@ -10,8 +12,10 @@ def __init__(self, fasta_file):
         self.faidx = {}
 
         self.fasta_file=fasta_file
+
         try:
-            self.fasta_handle=open(fasta_file)
+            self.fasta_fd = open(fasta_file)
+            self.fasta_handle = mmap.mmap(self.fasta_fd.fileno(), 0, access=mmap.ACCESS_READ)
         except IOError:
             print "Reference sequence doesn't exist"
 
@@ -83,6 +87,7 @@ def fetch_sequence(self, chrom, start, end):
 
     def __exit__(self, type, value, traceback):
         self.fasta_handle.close()
+        self.fasta_fd.close()
         self.faidx_handle.close()
 
     def chrm2len(self, chrm):