rstrip(*) from proteins if GFF/FASTA passed; #508 i was wrong!

nextgenusfs · Oct 30, 2020 · 05f0691 · 05f0691
1 parent d6c3899
commit 05f0691
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 18 deletions.
diff --git a/funannotate/annotate.py b/funannotate/annotate.py
@@ -434,12 +434,9 @@ def __init__(self, prog):
             else:
                 Scaffolds = args.fasta
                 GFF = args.gff
-                Proteins = os.path.join(
-                    outputdir, 'annotate_misc', 'genome.proteins.fa')
-                Transcripts = os.path.join(
-                    outputdir, 'annotate_misc', 'genome.transcripts.fasta')
-                annotTBL = os.path.join(
-                    outputdir, 'annotate_misc', 'genome.tbl')
+                Proteins = os.path.join(outputdir, 'annotate_misc', 'genome.proteins.fa')
+                Transcripts = os.path.join(outputdir, 'annotate_misc', 'genome.transcripts.fasta')
+                annotTBL = os.path.join(outputdir, 'annotate_misc', 'genome.tbl')
                 prefix = None
                 if args.rename:
                     prefix = args.rename.replace('_', '')
@@ -449,12 +446,9 @@ def __init__(self, prog):
                     GFF, prefix, Scaffolds, Proteins, Transcripts, annotTBL, external=True)
         else:
             genbank = args.genbank
-            Scaffolds = os.path.join(
-                outputdir, 'annotate_misc', 'genome.scaffolds.fasta')
-            Proteins = os.path.join(
-                outputdir, 'annotate_misc', 'genome.proteins.fasta')
-            Transcripts = os.path.join(
-                outputdir, 'annotate_misc', 'genome.transcripts.fasta')
+            Scaffolds = os.path.join(outputdir, 'annotate_misc', 'genome.scaffolds.fasta')
+            Proteins = os.path.join(outputdir, 'annotate_misc', 'genome.proteins.fasta')
+            Transcripts = os.path.join(outputdir, 'annotate_misc', 'genome.transcripts.fasta')
             GFF = os.path.join(outputdir, 'annotate_misc', 'genome.gff3')
             annotTBL = os.path.join(outputdir, 'annotate_misc', 'genome.tbl')
             lib.log.info("Checking GenBank file for annotation")
@@ -523,12 +517,9 @@ def __init__(self, prog):
                 lib.log.error(
                     "Found existing output directory %s. Warning, will re-use any intermediate files found." % (outputdir))
             lib.log.info("Parsing input files")
-            Scaffolds = os.path.join(
-                outputdir, 'annotate_misc', 'genome.scaffolds.fasta')
-            Proteins = os.path.join(
-                outputdir, 'annotate_misc', 'genome.proteins.fasta')
-            Transcripts = os.path.join(
-                outputdir, 'annotate_misc', 'genome.transcripts.fasta')
+            Scaffolds = os.path.join(outputdir, 'annotate_misc', 'genome.scaffolds.fasta')
+            Proteins = os.path.join(outputdir, 'annotate_misc', 'genome.proteins.fasta')
+            Transcripts = os.path.join(outputdir, 'annotate_misc', 'genome.transcripts.fasta')
             if TBL:
                 lib.log.info('Existing tbl found: {:}'.format(TBL))
                 shutil.copyfile(TBL, annotTBL)

diff --git a/funannotate/library.py b/funannotate/library.py
@@ -1994,9 +1994,13 @@ def _sortDict(d):
                         Transcript = str(v['transcript'][i])
                     except IndexError:
                         print((k, v))
+                    if v['strand'] == '-':
+                        Transcript = RevComp(Transcript)
                     tranout.write('>%s %s\n%s\n' % (x, k, softwrap(Transcript)))
                     if v['type'] == 'mRNA':
                         Prot = v['protein'][i]
+                        if Prot.endswith('*'):
+                            Prot = Prot.rstrip('*')
                         protout.write('>%s %s\n%s\n' % (x, k, softwrap(Prot)))
 
     return len(Genes), geneDB