Merge pull request #93 from Robaina/Robaina/issue71

Robaina/issue71
Robaina · Nov 3, 2023 · 5d7cfc9 · 5d7cfc9
2 parents a9c6d82 + 2573960
commit 5d7cfc9
Show file tree

Hide file tree

Showing 8 changed files with 390 additions and 124 deletions.
diff --git a/src/pynteny/cli.py b/src/pynteny/cli.py
@@ -517,6 +517,22 @@ def download() -> argparse.ArgumentParser:
             action="store_true",
             help="force-download database again if already downloaded",
         )
+        optional.add_argument(
+            "-pgap",
+            "--pgap",
+            dest="pgap",
+            default=False,
+            action="store_true",
+            help="download PGAP database (default)",
+        )
+        optional.add_argument(
+            "-pfam",
+            "--pfam",
+            dest="pfam",
+            default=False,
+            action="store_true",
+            help="download PFAM database",
+        )
         optional.add_argument(
             "-l",
             "--log",

diff --git a/src/pynteny/config.json b/src/pynteny/config.json
@@ -1,9 +1,11 @@
 {
     "database_dir": "",
     "upack_PGAP_database": false,
-    "data_downloaded": false,
+    "upack_PFAM_database": false,
+    "PGAP_data_downloaded": false,
+    "PFAM_data_downloaded": false,
     "PGAP_database": "",
     "PGAP_meta_file": "",
-    "streamlit_process": "",
-    "streamlit_log": ""
+    "PFAM_database": "",
+    "PFAM_meta_file": ""
 }
diff --git a/src/pynteny/filter.py b/src/pynteny/filter.py
@@ -16,7 +16,7 @@
 
 import pynteny.parsers.labelparser as labelparser
 import pynteny.parsers.syntenyparser as syntenyparser
-from pynteny.hmm import HMMER, PGAP
+from pynteny.hmm import HMMER, PFAM, PGAP, HMMDatabase
 from pynteny.preprocessing import FASTA
 
 logger = logging.getLogger(__name__)
@@ -416,14 +416,14 @@ def add_HMM_meta_info_to_hits(self, hmm_meta: Path) -> SyntenyHits:
         fields = ["gene_symbol", "label", "product", "ec_number"]
         if all([f in self._synteny_hits.columns for f in fields]):
             return self._synteny_hits
-        pgap = PGAP(hmm_meta)
+        hmm_db = HMMDatabase(hmm_meta)
         self._synteny_hits[fields] = ""
         for row in self._synteny_hits.itertuples():
             meta_values = [
                 [
                     str(v).replace("nan", "")
-                    for k, v in pgap.get_meta_info_for_HMM(hmm).items()
-                    if k != "#ncbi_accession"
+                    for k, v in hmm_db.get_meta_info_for_HMM(hmm).items()
+                    if k != "accession"
                 ]
                 for hmm in row.hmm.split("|")
             ]