gff config generate

config.yml

@@ -34,7 +34,8 @@ blastx:
 
 gff_format:
   genome_browser:
-    te_base: a
-    domain: b
-
+    te_base: nested_repeat
+    domain: polypeptide_domain
+    te: repeat_fragment
+    
 logdir: /tmp/nested/logs

nested/cli/nester.py

@@ -30,18 +30,19 @@ def main(input_fasta, sketch_only, format, data_folder):
             if not sketch_only:
                 nester = Nester(sequence)
                 sketcher.create_gff(nester.nested_element, format=format)
-            if format == 'default':
-                sketcher.sketch(sequence.id)
+                sketcher.create_gff(nester.nested_element)
+            
+            sketcher.sketch(sequence.id)
             seq_end_time = datetime.now()
             print('Processing {a}: DONE [{b}]'.format(a=sequence.id[:strlen], b=seq_end_time - seq_start_time)) 
         except KeyboardInterrupt:
             raise
-        #except CalledProcessError:
-        #    number_of_errors += 1
-        #    print('Processing {}: SUBPROCESS ERROR'.format(sequence.id[:strlen]))
-        #except:
-        #    number_of_errors += 1
-        #    print('Processing {}: UNEXPECTED ERROR:'.format(sequence.id[:strlen]), sys.exc_info()[0]) 
+        except CalledProcessError:
+            number_of_errors += 1
+            print('Processing {}: SUBPROCESS ERROR'.format(sequence.id[:strlen]))
+        except:
+            number_of_errors += 1
+            print('Processing {}: UNEXPECTED ERROR:'.format(sequence.id[:strlen]), sys.exc_info()[0]) 
 
 
     endTime = datetime.now()

nested/output/gff.py

@@ -69,13 +69,13 @@ class GFFMaker(object):
             cropped = intervals.crop(nl[i].location, children)
             for subinterval in cropped:
                 subseq += nested_element.sequence[subinterval[0]: (subinterval[1] + 1)]
-                if format == 'default':
-                    features.append(SeqFeature(
-                        FeatureLocation(subinterval[0], subinterval[1]),
-                        type='te',
-                        strand=0,
-                        qualifiers={'ID': 'TE {}'.format(i), 'Parent': 'TE_BASE {}'.format(i)}
-                    ))
+                te_type = format_dict[format]['te'] if format != 'default' else 'te'
+                features.append(SeqFeature(
+                    FeatureLocation(subinterval[0], subinterval[1]),
+                    type='te',
+                    strand=0,
+                    qualifiers={'ID': 'TE {}'.format(i), 'Parent': 'TE_BASE {}'.format(i)}
+                ))
 
             # save transposon fasta
             with open('{}/{}/TE/{}.fa'.format(dirpath, nested_element.id, i), 'w') as fasta_out:
@@ -108,7 +108,10 @@ class GFFMaker(object):
         rec.features = features
 
         #create GFF
-        gff_filepath = '{}/{}/{}.gff'.format(dirpath, nested_element.id, nested_element.id)
+        filename = '{}/{}/{}'.format(dirpath, nested_element.id, nested_element.id)
+        if format != 'default':
+            filename += '_{}'.format(format)
+        gff_filepath = '{}.gff'.format(filename)
         with open(gff_filepath, 'w+') as gff_out:
             GFF.write([rec], gff_out)