add new parameter to set source string value in gtf and gff second column to ease file merging

miniannotator.py

@@ -367,7 +367,7 @@ class Miniannotator:
 
         return gene_id, gene_o
 
-    def write_gtf_file(self, gtf_file, genes):
+    def write_gtf_file(self, gtf_file, genes, source):
         """
         Write genes, transcripts, exons and indels to a GTF file
 
@@ -380,7 +380,7 @@ class Miniannotator:
 
         with open(gtf_file, "w") as gtf:
             for gene_id, gene in genes.items():
-                line = '{seqname}\tminiannotator\t{feature}\t{start}\t{end}\t.\t{strand}\t.\tgene_id "{gene_id}"' \
+                line = '{seqname}\t'+source+'\t{feature}\t{start}\t{end}\t.\t{strand}\t.\tgene_id "{gene_id}"' \
                        '{attrs}\n'
                 gtf.write(line.format(
                     seqname=gene["seqname"],
@@ -430,7 +430,7 @@ class Miniannotator:
                             attrs=attrs_ex
                         ))
 
-    def write_gff_file(self, gff_file, genes):
+    def write_gff_file(self, gff_file, genes, source):
         """
         Write genes, transcripts, exons and indels to a GFF file
 
@@ -443,7 +443,7 @@ class Miniannotator:
 
         with open(gff_file, "w") as gff:
             for gene_id, gene in genes.items():
-                line = '{seqname}\tminiannotator\t{feature}\t{start}\t{end}\t.\t{strand}\t.\t{attrs}\n'     
+                line = '{seqname}\t'+source+'\t{feature}\t{start}\t{end}\t.\t{strand}\t.\t{attrs}\n'     
                 attrs = 'ID={gene_id}'.format(gene_id=gene_id)
                 gff.write(line.format(
                     seqname=gene["seqname"],
@@ -481,7 +481,7 @@ class Miniannotator:
 
            
 
-    def search_genes(self, gtf_file):
+    def search_genes(self, gtf_file, source):
         """
         Parse BAM file to search genes and exons positions
         Query match position on the reference defines gene position
@@ -522,11 +522,11 @@ class Miniannotator:
         full_genes_2 = OrderedDict(pool.map(Miniannotator._rename_exons, list(full_genes.items())))
 
         print("Writing to GTF file...", flush=True)
-        self.write_gtf_file(gtf_file=gtf_file, genes=full_genes_2)
+        self.write_gtf_file(gtf_file=gtf_file, genes=full_genes_2, source=source)
 
         print("Writing to GFF file...", flush=True)
         gff_file = os.path.splitext(gtf_file)[0]+".gff"
-        self.write_gff_file(gff_file=gff_file, genes=full_genes_2)
+        self.write_gff_file(gff_file=gff_file, genes=full_genes_2, source=source)
 
 if __name__ == "__main__":
     import argparse
@@ -539,6 +539,7 @@ if __name__ == "__main__":
     parser.add_argument("-q", "--min-qoverlap", help="Minimal query overlap [0-100]", type=int, default=90,
                         required=False)
     parser.add_argument("-o", "--output-dir", help="Output folder path", required=False, default=".")
+    parser.add_argument("-s", "--source", help="Source description string (second column in gtf or gff files)", required=False, default="miniannotator")
 
     args = parser.parse_args()
 
@@ -568,4 +569,4 @@ if __name__ == "__main__":
                                   map=args.map)
 
     # Search genes
-    annotator.search_genes(os.path.join(args.output_dir, "annotations.gtf"))
+    annotator.search_genes(os.path.join(args.output_dir, "annotations.gtf"), args.source)