From c44ea033649ef5027f3b0ea727c0462468cde25e Mon Sep 17 00:00:00 2001
From: Celine Noirot <celine.noirot@inra.fr>
Date: Mon, 27 Feb 2017 17:45:12 +0000
Subject: [PATCH] Get species from swissprot definition Force creation of
symbolic link in addinstance.
---
workflows/addinstance/__init__.py | 2 +-
workflows/rnaseqdenovo/bin/blast2annot.py | 24 ++++++++---------------
2 files changed, 9 insertions(+), 17 deletions(-)
diff --git a/workflows/addinstance/__init__.py b/workflows/addinstance/__init__.py
index 609c7b8b..803b018e 100644
--- a/workflows/addinstance/__init__.py
+++ b/workflows/addinstance/__init__.py
@@ -86,7 +86,7 @@ class AddInstance (Workflow):
subprocess.check_call( self.get_exec_path("sed") + " -i -e 's,^session\.connection.*,session.connection = jdbc:mysql://{host}:{port}/biomart_sessions?user={user}\&password={passwd},' {propfile}".format(
propfile = os.path.join( NGSPIPELINES_DIST, 'biomart.properties'), **ngs_reader.get_database_info() ), shell = True)
#create link data
- subprocess.check_call( self.get_exec_path("ln") + ' -s ' + os.path.join( instance_directory , 'data') + ' ' + os.path.join( NGSPIPELINES_DIST_PLUGIN, 'ngspipelines', 'public','ngspipelines' ) , shell = True)
+ subprocess.check_call( self.get_exec_path("ln") + ' -f -s ' + os.path.join( instance_directory , 'data') + ' ' + os.path.join( NGSPIPELINES_DIST_PLUGIN, 'ngspipelines', 'public','ngspipelines' ) , shell = True)
if self.instance_url :
url = re.sub(r':\d+$', '', re.sub( r'\/$', '', str(self.instance_url))) + ':' + str(self.instance_port)
if not url.startswith("http") :
diff --git a/workflows/rnaseqdenovo/bin/blast2annot.py b/workflows/rnaseqdenovo/bin/blast2annot.py
index 2ea418d9..f61f8351 100755
--- a/workflows/rnaseqdenovo/bin/blast2annot.py
+++ b/workflows/rnaseqdenovo/bin/blast2annot.py
@@ -62,15 +62,7 @@ def spFromHit( alignment ):
@param alignment : [Bio::blast::Record::Alignment] the alignment.
"""
species = None
- # GI
- if re.match("gi\|\d+\|[^\|]+\|.+", alignment.hit_id) is not None:
- try:
- matches = re.search('\[([^\]]+)\]', alignment.hit_def)
- species = matches.group(1)
- except: # swissprot
- pass
- # GNL
- elif re.match('gnl\|\w+\|\d+', alignment.hit_id) is not None:
+ if re.match(".*OS=.*", alignment.hit_def) is not None:
try:
matches = re.search("OS=([^=]+)", alignment.hit_def)
if matches is not None:
@@ -79,13 +71,13 @@ def spFromHit( alignment ):
clean_matches = re.search("(.+) [A-Z][A-Z]$", species)
if clean_matches is not None:
species = clean_matches.group(1)
- else:
- matches = re.search('^gi\|\d+\|', alignment.hit_def)
- if matches is not None:
- cross_references = re.split("gi\|\d+\|", alignment.hit_def )
- matches = re.search( '\[([^\]]+)\][^\[]*$', cross_references[1] )
- species = matches.group(1)
- except: # ensembl
+ except:
+ pass
+ else :
+ try:
+ matches = re.search('\[([^\]]+)\]', alignment.hit_def)
+ species = matches.group(1)
+ except:
species = None
pass
# Others databases are not implemented
--
GitLab