fixups for vcf_filter

scripts/vcf_filter.py

@@ -17,13 +17,13 @@
 parser.add_argument('filters', metavar='filter', type=str, nargs='*', default=None,
         help='Filters to use')
 parser.add_argument('--no-short-circuit', action='store_true',
-        help='Do not stop filter processing on a site if a single filter fails.')
+        help='Do not stop filter processing on a site if a single filter fails')
 parser.add_argument('--output', action='store', default=sys.stdout,
-        help='Filename to output (default stdout)')
+        help='Filename to output [stdout]')
 parser.add_argument('--no-filtered', action='store_true',
-        help='Remove failed sites')
+        help='Output only sites passing the filters')
 parser.add_argument('--local-script', action='store', default=None,
-        help='File in current working directory with filter classes (with .py)')
+        help='Python file in current working directory with the filter classes')
 
 
     # TODO: allow filter specification by short name
@@ -35,38 +35,44 @@
 def main():
     # dynamically build the list of available filters
     filters = {}
-    filter_help = '\n\navailable filters:'
+    filter_help = ['\navailable filters:']
 
     def addfilt(filt, help):
         filters[filt.name] = filt
         filt.customize_parser(parser)
-        help += '\n  %s:\t%s' % (filt.name, filt.description)
+        help.append('  %s:\t%s' % (filt.name, filt.description))
 
     for p in pkg_resources.iter_entry_points('vcf.filters'):
         filt = p.load()
         addfilt(filt, filter_help)
 
     # parse command line args
+    # (mainly because of local_script)
     args = parser.parse_args()
 
     # add all classes from local script, if present
     if args.local_script != None:
         import inspect
+        import sys, os
+        sys.path.insert(0, os.getcwd())
         module_name = args.local_script.replace('.py', '')
         mod = __import__(module_name)
         classes = inspect.getmembers(mod, inspect.isclass)
         for name, cls in classes:
             addfilt(cls, filter_help)
 
-    parser.description += filter_help
-
+        # apply the updated argument definitions
+        args = parser.parse_args()
+
+    parser.description += '\n'.join(filter_help)
+
     # show help after loading of the local module
     # so it can include the additional options
     if args.help or len(args.filters) == 0 or args.input == None:
         parser.print_help()
         parser.exit()
 
-    inp = vcf.Reader(file(args.input[0]))
+    inp = vcf.Reader(file(args.input))
 
     # build filter chain
     chain = []
@@ -79,16 +85,20 @@ def addfilt(filt, help):
 
     # apply filters
     short_circuit = not args.no_short_circuit
+    drop_filtered = args.no_filtered
 
     for record in inp:
         for filt in chain:
             result = filt(record)
-            if result:
+            if result != None:
+                # save some work by skipping the rest of the code
+                if drop_filtered: break
+
                 record.add_filter(filt.filter_name())
-                if short_circuit:
-                    break
-
-        if (not args.no_filtered) or (record.FILTER == '.'):
+                if short_circuit: break
+        else:
+            # use PASS only if other filter names appear in the FILTER column
+            if record.FILTER == '.' and not drop_filtered: record.FILTER = 'PASS'
             oup.write_record(record)
 
 if __name__ == '__main__': main()

vcf/parser.py

@@ -255,7 +255,9 @@ def add_format(self, fmt):
         self.FORMAT = self.FORMAT + ':' + fmt
 
     def add_filter(self, flt):
-        if self.FILTER is None or self.FILTER == 'PASS':
+        if self.FILTER is None \
+        or self.FILTER == 'PASS'\
+        or self.FILTER == '.':
             self.FILTER = ''
         else:
             self.FILTER = self.FILTER + ';'