fetchfa: Avoid HTTP 414 error; blast2accmap: Print gi if exists

jlhg · Jan 30, 2013 · 24e189a · 24e189a
1 parent a1b4c0c
commit 24e189a
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 8 deletions.
diff --git a/bdorpy/blast2accmap.py b/bdorpy/blast2accmap.py
@@ -7,7 +7,7 @@
 # http://opensource.org/licenses/MIT
 #
 # Author: Jian-Long Huang (jianlong@ntu.edu.tw)
-# Version: 0.4
+# Version: 0.5
 # Created: 2013.1.25
 #
 # Required:
@@ -29,6 +29,7 @@
 
 import sys
 import argparse
+import re
 from Bio.Blast import NCBIXML
 from fhandle import name, logmsg
 
@@ -63,9 +64,11 @@ def main():
         fw.write('# E-value threshold: ' + str(args.ev_thresh) + '\n')
         fw.write('# min hit number: ' + str(args.min_hit_num) + '\n')
         fw.write('#\n')
-        fw.write('# query_accession    hit_accession_1,hit_accession_2, ...\n\n')
+        fw.write('# filename    query_accession,hit_accession_1,hit_accession_2, ...\n\n')
         fw.flush()
 
+        gi = re.compile('gi\|(\d+)\|')
+
         for blast_record in blast_records:
             total_query_num += 1
 
@@ -81,7 +84,13 @@ def main():
                         continue
 
                     if hsp.expect <= args.ev_thresh:
-                        hit_accs.append(alignment.accession)
+                        match = gi.match(alignment.hit_id).group(1)
+
+                        if match is None:
+                            print(alignment.accession + ' does not have gi.')
+                            hit_accs.append(alignment.accession)
+                        else:
+                            hit_accs.append(match)
                         break
 
             if len(hit_accs) >= args.min_hit_num:

diff --git a/bdorpy/fetchfa.py b/bdorpy/fetchfa.py
@@ -7,7 +7,7 @@
 # http://opensource.org/licenses/MIT
 #
 # Author: Jian-Long Huang (jianlong@ntu.edu.tw)
-# Version: 0.3
+# Version: 0.4
 # Created: 2013.1.24
 #
 # Required:
@@ -59,7 +59,7 @@ def main():
         fwlog.write(i)
     fwlog.flush()
 
-    Entrez.email = 'fetchfa@example.com'
+    Entrez.email = name.genid() + '@example.com'
 
     if args.query_id is not None:
         with open(args.output + '.fa', 'w') as fw, open(args.output + '.log', 'w') as fwlog:
@@ -90,11 +90,28 @@ def main():
                 query_num += 1
 
                 with open(os.path.abspath(args.output) + '/' + line.split('\t')[0] + '.fa', 'w') as fw:
+                    alist = line.rstrip().split('\t')[1].split(',')
+
+                    while len(alist) > 30:
+                        alist_part = alist[0:30]
+                        alist = alist[30:len(alist)]
+
+                        handle = Entrez.efetch(db=args.database,
+                                               id=','.join(alist_part),
+                                               rettype='fasta',
+                                               retmode='text')
+                        fw.write(handle.read())
+                        fw.flush()
+                        handle.close()
+
                     handle = Entrez.efetch(db=args.database,
-                                           id=line.split('\t')[0] + ',' + line.split('\t')[1].rstrip(),
+                                           id=','.join(alist),
                                            rettype='fasta',
                                            retmode='text')
                     fw.write(handle.read())
+                    fw.flush()
+                    handle.close()
+
             fwlog.write('# Fetched queries: ' + str(query_num) + '\n')
             fwlog.write('#\n')
 

diff --git a/docs/format/blastaccmap.txt b/docs/format/blastaccmap.txt
@@ -3,5 +3,5 @@
 # Updated: 2013.1.25
 
 col_index    title
-0            query_accession
-1            hit_accession_1,hit_accession_2, ...
+0            output file name to be created
+1            accession_1,accession_2, ...