Merge pull request #6 from MPI-Dortmund/tsv-as-argument

AM database as argument

Author: thorstenwagner

README.md

@@ -29,7 +29,7 @@ pip install pymissense
 
 Generate usage is:
 ```
-usage: pymissense [-h] [--pdbpath PDBPATH] [--maxacid MAXACID] uniprot_id output_path
+usage: pymissense[-h] [--tsv TSV] [--pdbpath PDBPATH] [--maxacid MAXACID] uniprot_id output_path
 
 AlphaMissense plot and pdb generator
 
@@ -39,13 +39,14 @@ positional arguments:
 
 options:
   -h, --help         show this help message and exit
+  --tsv TSV          You can provide the path to the tsv file if you want to skip the download. (default: None)
   --pdbpath PDBPATH  If defined, it will write the pathogencity as bfactor in that PDB. If its not defined or not existing it will instead download the alphafold predicted PDB (default: None)
-  --maxacid MAXACID  Maximum squence number to use. (default: None)
+  --maxacid MAXACID  Maximum squence number to use in the plot. (default: None)
 ```
 
 You can provide the optional argument `--pdbpath` if you want to use an experimental PDB, otherwise it will instead download the alphafold predicted PDB.
 
-For example, to reproduce [Figure 3D](https://www.science.org/doi/10.1126/science.adg7492#F3) (the middle one) and the generate the PDB shown in [Figure 3E](https://www.science.org/doi/10.1126/science.adg7492#F3) do:
+For example, to reproduce [Figure 3D](https://www.science.org/doi/10.1126/science.adg7492#F3) (the middle one) and generates the PDB shown in [Figure 3E](https://www.science.org/doi/10.1126/science.adg7492#F3) do:
 
 ```
 wget https://files.rcsb.org/download/7UPI.pdb

missense/missense.py

@@ -401,11 +401,13 @@ def gen_image(pos_to_val) -> np.array:
     return img
 
 
-def get_data_tuple(uniprot_id: str):
+def get_data_tuple(uniprot_id: str, tsv_path: str = None):
     """
     Extracts the raw data for the plot from the tsv file.
     """
-    with open(os.path.join(tempfile.gettempdir(), "alpha.tsv"), encoding="utf-8") as f:
+    if tsv_path is None:
+        tsv_path = os.path.join(tempfile.gettempdir(), "alpha.tsv")
+    with open(tsv_path, encoding="utf-8") as f:
         doc = f.read()
         m = re.findall(uniprot_id.upper() + r"\t(.\d+.)\t(\d.\d+)", doc)
     pos_to_val = []
@@ -457,6 +459,13 @@ def create_parser() -> argparse.ArgumentParser:
         help="Output folder",
     )
 
+    parser.add_argument(
+        "--tsv",
+        type=str,
+        help="You can provide the path to the tsv file if you want to skip the download.",
+        default=None
+    )
+
     parser.add_argument(
         "--pdbpath",
         type=str,
@@ -547,8 +556,17 @@ def create_modified_pdb(img: np.array, uniprot_id: str, output_path: str, pdb_pt
                 else:
                     out_file.write(f'{line}')
 
-def _run(uniprot_id: str, output_path: str, pdbpath: str, maxacid: int):
-    download_missense_data()
+def _run(uniprot_id: str,
+         output_path: str,
+         tsvpath: str,
+         pdbpath: str,
+         maxacid: int):
+
+    if tsvpath is None or os.path.exists(tsvpath)==False:
+        tsvpath=None
+        download_missense_data()
+
+
     os.makedirs(output_path, exist_ok=True)
 
     chain = None
@@ -559,7 +577,11 @@ def _run(uniprot_id: str, output_path: str, pdbpath: str, maxacid: int):
             print(f"Cant find chain for {uniprot_id} in {pdbpath}")
             sys.exit(1)
 
-    pos_to_val = get_data_tuple(uniprot_id)
+    pos_to_val = get_data_tuple(uniprot_id=uniprot_id, tsv_path=tsvpath)
+
+    if len(pos_to_val) == 0:
+        print(f"Could not find any data in the AlphaMissense database for uniprot id {uniprot_id}")
+        sys.exit(1)
 
     out_fig_pth = os.path.join(output_path, f"{uniprot_id}.pdf")
     img_raw_data = make_and_save_plot(pos_to_val, out_fig_pth, maxacid)
@@ -570,7 +592,11 @@ def _run(uniprot_id: str, output_path: str, pdbpath: str, maxacid: int):
 
 def _main_():
     args = create_parser().parse_args()
-    _run(args.uniprot_id, args.output_path, args.pdbpath, args.maxacid)
+    _run(uniprot_id=args.uniprot_id,
+         output_path=args.output_path,
+         pdbpath=args.pdbpath,
+         maxacid=args.maxacid,
+         tsvpath=args.tsv)
 
 
 

tests/test_run.py

@@ -28,7 +28,35 @@ def test_pdf_and_and_are_generated(self):
             ms._run(uniprot_id="Q9UQ13",
                     output_path=tmpdirname,
                     maxacid=200,
-                    pdbpath=None
+                    pdbpath=None,
+                    tsvpath=None
+                    )
+
+            self.assertEqual(True, os.path.exists(os.path.join(tmpdirname,"Q9UQ13.pdf")))
+            self.assertEqual(True, os.path.exists(os.path.join(tmpdirname, "Q9UQ13-edit.pdb")))
+
+    def test_pdf_and_and_are_generated_non_existing_uniprot(self):
+
+        shutil.copyfile(os.path.join(os.path.dirname(__file__), "../resources/tests/Q9UQ13/alpha.tsv"),
+                        os.path.join(tempfile.gettempdir(), "alpha.tsv"))
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            with self.assertRaises(SystemExit):
+                ms._run(uniprot_id="Q9UQ13234234",
+                        output_path=tmpdirname,
+                        maxacid=200,
+                        pdbpath=None,
+                        tsvpath=None
+                        )
+
+    def test_pdf_and_and_are_generated_tsv_by_path(self):
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            ms._run(uniprot_id="Q9UQ13",
+                    output_path=tmpdirname,
+                    maxacid=200,
+                    pdbpath=None,
+                    tsvpath=os.path.join(os.path.dirname(__file__), "../resources/tests/Q9UQ13/alpha.tsv")
                     )
 
             self.assertEqual(True, os.path.exists(os.path.join(tmpdirname,"Q9UQ13.pdf")))
@@ -43,7 +71,8 @@ def test_pdb_check_with_reference(self):
             ms._run(uniprot_id="Q9UQ13",
                     output_path=tmpdirname,
                     maxacid=200,
-                    pdbpath=None
+                    pdbpath=None,
+                    tsvpath=None
                     )
 
             ref_pth = os.path.join(os.path.dirname(__file__), "../resources/tests/Q9UQ13/Q9UQ13-edit.pdb")
@@ -62,7 +91,8 @@ def test_pdb_check_with_reference_with_pdb(self):
             ms._run(uniprot_id="Q9UQ13",
                     output_path=tmpdirname,
                     maxacid=200,
-                    pdbpath=os.path.join(os.path.dirname(__file__), "../resources/tests/Q9UQ13-with-pdb/7upi.pdb")
+                    pdbpath=os.path.join(os.path.dirname(__file__), "../resources/tests/Q9UQ13-with-pdb/7upi.pdb"),
+                    tsvpath=None
                     )
 
             ref_pth = os.path.join(os.path.dirname(__file__), "../resources/tests/Q9UQ13-with-pdb/Q9UQ13-edit.pdb")