From 16b2e987ae4cd065ddd6b9885d4ae81f6a533dbf Mon Sep 17 00:00:00 2001
From: "Jinlong Ru@hms" <jinlong.ru@helmholtz-muenchen.de>
Date: Sun, 4 Sep 2022 20:08:12 +0200
Subject: [PATCH] add feature_prefix

---
 bin/add_prophage_to_gbk.py  | 38 +------------------------------------
 bin/extract_CDS_from_gbk.py | 13 +++++++------
 2 files changed, 8 insertions(+), 43 deletions(-)
 mode change 100755 => 120000 bin/add_prophage_to_gbk.py

diff --git a/bin/add_prophage_to_gbk.py b/bin/add_prophage_to_gbk.py
deleted file mode 100755
index 7b0e357..0000000
--- a/bin/add_prophage_to_gbk.py
+++ /dev/null
@@ -1,37 +0,0 @@
-#!/usr/bin/env python
-
-import click
-from Bio import SeqFeature, SeqIO
-import pandas as pd
-
-
-def add_prophage(record, prophage_loc):
-    start_pos = SeqFeature.ExactPosition(prophage_loc['start'])
-    end_pos = SeqFeature.ExactPosition(prophage_loc['end'])
-    feature_location = SeqFeature.FeatureLocation(start_pos-1, end_pos, strand=1)
-    qualifiers = {"ID": prophage_loc['prophage_id'], "note": "Prophage"}
-
-    new_feature = SeqFeature.SeqFeature(feature_location, type="misc_feature", qualifiers=qualifiers)
-    record.features.append(new_feature)
-    return record
-
-
-@click.command()
-@click.option("--gbk", '-g', help="input file")
-@click.option("--prophage_coords", '-p', help="Prophage coordinates file")
-@click.option("--fout", '-o', help="output file name")
-def main(gbk, prophage_coords, fout):
-    records = list(SeqIO.parse(gbk, format="genbank"))
-    prophages = pd.read_csv(prophage_coords, sep='\t', names = ['ref_id', 'start', 'end', 'prophage_id']).to_dict(orient="records")
-    
-    for rec in records:
-        for pg_loc in prophages:
-            if rec.id == pg_loc['ref_id']:
-                rec = add_prophage(rec, pg_loc)
-    
-    with open(fout, "w") as fh:
-        SeqIO.write(records, fh, "genbank")
-
-
-if __name__ == '__main__':
-    main()
\ No newline at end of file
diff --git a/bin/add_prophage_to_gbk.py b/bin/add_prophage_to_gbk.py
new file mode 120000
index 0000000..5dfc829
--- /dev/null
+++ b/bin/add_prophage_to_gbk.py
@@ -0,0 +1 @@
+/home/viro/jinlong.ru/github/VIRO_pipelines/script/add_prophage_to_gbk.py
\ No newline at end of file
diff --git a/bin/extract_CDS_from_gbk.py b/bin/extract_CDS_from_gbk.py
index a884daf..d6e9b54 100755
--- a/bin/extract_CDS_from_gbk.py
+++ b/bin/extract_CDS_from_gbk.py
@@ -4,7 +4,7 @@
 from Bio import SeqIO, SeqRecord, Seq
 
 
-def extract_CDSs(gbks, fout_DNA, fout_AA):
+def extract_CDSs(gbks, fout_DNA, fout_AA, feature_prefix):
     cds_DNAs = []
     cds_AAs = []
     for gbk in gbks:
@@ -15,11 +15,11 @@ def extract_CDSs(gbks, fout_DNA, fout_AA):
             protein_seq = cds_feature.qualifiers.get('translation')
             if protein_seq:
                 protein_seq = protein_seq[0]
-                cds_AA = SeqRecord.SeqRecord(seq=Seq.Seq(protein_seq), id=locus_tag, description="")
+                cds_AA = SeqRecord.SeqRecord(seq=Seq.Seq(protein_seq), id="{}{}".format(feature_prefix, locus_tag), description="")
                 cds_AAs.append(cds_AA)
             
                 cds_DNA =  cds_feature.location.extract(gbk)
-                cds_DNA.id=locus_tag
+                cds_DNA.id="{}{}".format(feature_prefix, locus_tag)
                 cds_DNA.description = ""
                 cds_DNAs.append(cds_DNA)
         
@@ -38,7 +38,8 @@ def extract_genome(gbks, fout):
 @click.command()
 @click.option("--fgbk", '-i', help="Genbank file")
 @click.option("--fout_prefix", '-o', help="Genbank file")
-def main(fgbk, fout_prefix):
+@click.option("--feature_prefix", '-p', default="", required=False, help="Genbank file")
+def main(fgbk, fout_prefix, feature_prefix):
     gbks = list(SeqIO.parse(fgbk, format="genbank"))
 
     # Output
@@ -47,8 +48,8 @@ def main(fgbk, fout_prefix):
     fout_cds_AA = "{}_protein.faa".format(fout_prefix)
 
     extract_genome(gbks, fout_genome)
-    extract_CDSs(gbks, fout_cds_DNA, fout_cds_AA)
+    extract_CDSs(gbks, fout_cds_DNA, fout_cds_AA, feature_prefix)
 
 
 if __name__ == '__main__':
-    main()
\ No newline at end of file
+    main()