From 586b7afe830cb3f4c6970ad79278db4726e11192 Mon Sep 17 00:00:00 2001
From: Oliver Schwengers <oliver.schwengers@computational.bio.uni-giessen.de>
Date: Thu, 9 Mar 2023 09:10:17 +0100
Subject: [PATCH] add DB type info to outputs #95

---
 bakta/io/gff.py   | 2 +-
 bakta/io/insdc.py | 5 +----
 bakta/io/json.py  | 5 ++++-
 bakta/io/tsv.py   | 8 ++++++--
 bakta/main.py     | 2 +-
 5 files changed, 13 insertions(+), 9 deletions(-)

diff --git a/bakta/io/gff.py b/bakta/io/gff.py
index 397e49ca..4af60252 100644
--- a/bakta/io/gff.py
+++ b/bakta/io/gff.py
@@ -27,7 +27,7 @@ def write_gff3(genome: dict, features_by_contig: Dict[str, dict], gff3_path: Pat
 
         fh.write('# Annotated with Bakta\n')
         fh.write(f'# Software: v{bakta.__version__}\n')
-        fh.write(f"# Database: v{cfg.db_info['major']}.{cfg.db_info['minor']}\n")
+        fh.write(f"# Database: v{cfg.db_info['major']}.{cfg.db_info['minor']}, {cfg.db_info['type']}\n")
         fh.write(f'# DOI: {bc.BAKTA_DOI}\n')
         fh.write(f'# URL: {bc.BAKTA_URL}\n')
 
diff --git a/bakta/io/insdc.py b/bakta/io/insdc.py
index 20f3f02b..73280aa6 100644
--- a/bakta/io/insdc.py
+++ b/bakta/io/insdc.py
@@ -27,15 +27,12 @@ def write_insdc(genome: dict, features: Sequence[dict], genbank_output_path: Pat
         comment = (
             'Annotated with Bakta',
             f"Software: v{bakta.__version__}\n",
-            f"Database: v{cfg.db_info['major']}.{cfg.db_info['minor']}\n",
+            f"Database: v{cfg.db_info['major']}.{cfg.db_info['minor']}, {cfg.db_info['type']}\n",
             f'DOI: {bc.BAKTA_DOI}\n',
             f'URL: {bc.BAKTA_URL}\n',
             '\n',
             '##Genome Annotation Summary:##\n',
             f"{'Annotation Date':<30} :: {datetime.now().strftime('%m/%d/%Y, %H:%M:%S')}\n",
-            f"{'Annotation Pipeline':<30} :: Bakta\n",
-            f"{'Annotation Software version':<30} ::  v{bakta.__version__}\n",
-            f"{'Annotation Database version':<30} ::  v{cfg.db_info['major']}.{cfg.db_info['minor']}\n",
             f"{'CDSs':<30} :: {len([feat for feat in contig_features if feat['type'] == bc.FEATURE_CDS or feat['type'] == bc.FEATURE_SORF]):5,}\n",
             f"{'tRNAs':<30} :: {len([feat for feat in contig_features if feat['type'] == bc.FEATURE_T_RNA]):5,}\n",
             f"{'tmRNAs':<30} :: {len([feat for feat in contig_features if feat['type'] == bc.FEATURE_TM_RNA]):5,}\n",
diff --git a/bakta/io/json.py b/bakta/io/json.py
index bddaa934..afc35e60 100644
--- a/bakta/io/json.py
+++ b/bakta/io/json.py
@@ -63,7 +63,10 @@ def write_json(genome: dict, features: Sequence[dict], json_path: Path):
 
     version = OrderedDict()
     version['bakta'] = bakta.__version__
-    version['db'] = f"{cfg.db_info['major']}.{cfg.db_info['minor']}"
+    version['db'] = {
+        'version': f"{cfg.db_info['major']}.{cfg.db_info['minor']}",
+        'type': cfg.db_info['type']
+    }
     output['version'] = version
 
     with json_path.open('wt') as fh:
diff --git a/bakta/io/tsv.py b/bakta/io/tsv.py
index ca991498..bfe68e07 100644
--- a/bakta/io/tsv.py
+++ b/bakta/io/tsv.py
@@ -17,9 +17,13 @@ def write_tsv(contigs: Sequence[dict], features_by_contig: Dict[str, dict], tsv_
     log.info('write tsv: path=%s', tsv_path)
 
     with tsv_path.open('wt') as fh:
-        fh.write(f'#Annotated with Bakta (v{bakta.__version__}): https://github.com/oschwengers/bakta\n')
-        fh.write(f"#Database (v{cfg.db_info['major']}.{cfg.db_info['minor']}): https://doi.org/10.5281/zenodo.4247252\n")
+        fh.write('# Annotated with Bakta\n')
+        fh.write(f'# Software: v{bakta.__version__}\n')
+        fh.write(f"# Database: v{cfg.db_info['major']}.{cfg.db_info['minor']}, {cfg.db_info['type']}\n")
+        fh.write(f'# DOI: {bc.BAKTA_DOI}\n')
+        fh.write(f'# URL: {bc.BAKTA_URL}\n')
         fh.write('#Sequence Id\tType\tStart\tStop\tStrand\tLocus Tag\tGene\tProduct\tDbXrefs\n')
+
         for contig in contigs:
             for feat in features_by_contig[contig['id']]:
                 feat_type = feat['type']
diff --git a/bakta/main.py b/bakta/main.py
index 480ec8de..021ce214 100755
--- a/bakta/main.py
+++ b/bakta/main.py
@@ -601,7 +601,7 @@ def main():
         fh_out.write(f"oriTs: {len([f for f in features if f['type'] == bc.FEATURE_ORIT])}\n")
         fh_out.write('\nBakta:\n')
         fh_out.write(f'Software: v{bakta.__version__}\n')
-        fh_out.write(f"Database: v{cfg.db_info['major']}.{cfg.db_info['minor']}\n")
+        fh_out.write(f"Database: v{cfg.db_info['major']}.{cfg.db_info['minor']}, {cfg.db_info['type']}\n")
         fh_out.write('DOI: 10.1099/mgen.0.000685\n')
         fh_out.write('URL: github.com/oschwengers/bakta\n')