RNApolis 0.3.15__py3-none-any.whl → 0.3.17__py3-none-any.whl

{RNApolis-0.3.15.dist-info → RNApolis-0.3.17.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: RNApolis
-Version: 0.3.15
+Version: 0.3.17
 Summary: A Python library containing RNA-related bioinformatics functions and classes
 Home-page: https://github.com/tzok/rnapolis-py
 Author: Tomasz Zok

{RNApolis-0.3.15.dist-info → RNApolis-0.3.17.dist-info}/RECORD

@@ -1,17 +1,17 @@
 rnapolis/annotator.py,sha256=XnjFBeu3P_2UMdkD4Ao7m7K6JfeqYa-13xRzghrLvt8,22086
 rnapolis/clashfinder.py,sha256=jD3s_UovygWi01NUbQNeAeRRFkARTSRraLXUV43UbAA,8514
 rnapolis/common.py,sha256=tbmKV64nLClmzNN0HyKnvWA0-5aSyf4x8CGpBjpIJ4M,30290
-rnapolis/metareader.py,sha256=4qtMKRvww2sUStLeV8WVrLEt-ScydHUv4Gxx96tnf-M,1683
+rnapolis/metareader.py,sha256=9HZIrpPsmmWkEUeSeKAeG1sstfqbBDKUw5QPsn5cshw,2072
 rnapolis/molecule_filter.py,sha256=NhjuqdCRnXgPefWZPeTq77tifmnAzamQtA0ODqPPG9k,6918
 rnapolis/motif_extractor.py,sha256=duHvpi9Ulcny9K60E6VBpz5RpJZw-KdTB4_Ph0iP478,774
-rnapolis/parser.py,sha256=Cmjt7p8UkiSNhSQDjc6I7BRqtuIIHs23Fp-Glb8Zikw,12216
+rnapolis/parser.py,sha256=nHRXcoKbzfMJuDMJO-8MSPkzfmD-788ZzgcfjLQpgFs,12666
 rnapolis/rfam_folder.py,sha256=3rgXEJR16uPFy_BOo8qkdClOAOQDVOkidnLE-yoRbeI,11112
 rnapolis/tertiary.py,sha256=VuATTN2SD7lBL9iUgT-doDwuEYsLodgV2u-SwQsyQcU,19658
 rnapolis/transformer.py,sha256=V9nOQvdq4-p7yUWo0vQg0CDQMpmyxz9t4TMSRVEKHnw,1817
 rnapolis/util.py,sha256=IdquFO3PV1_KDqodjupzm0Rqvgy0CeSzxGHaGEHYXVU,543
-RNApolis-0.3.15.dist-info/LICENSE,sha256=ZGRu12MzCgbYA-Lt8MyBlmjvPZh7xfiD5u5wBx0enq4,1066
-RNApolis-0.3.15.dist-info/METADATA,sha256=JINDTFBZOiXrJUB3-jnBW8YIHcRUc3mQ2EKoaFDS5zE,54301
-RNApolis-0.3.15.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-RNApolis-0.3.15.dist-info/entry_points.txt,sha256=foN2Pn5e-OzEz0fFmNoX6PnFSZFQntOlY8LbognP5F0,308
-RNApolis-0.3.15.dist-info/top_level.txt,sha256=LcO18koxZcWoJ21KDRRRo_tyIbmXL5z61dPitZpy8yc,9
-RNApolis-0.3.15.dist-info/RECORD,,
+RNApolis-0.3.17.dist-info/LICENSE,sha256=ZGRu12MzCgbYA-Lt8MyBlmjvPZh7xfiD5u5wBx0enq4,1066
+RNApolis-0.3.17.dist-info/METADATA,sha256=2h_8NfBq0uKpgXC-CYsvB1O1hglxTGyZHfGfJmsKrmI,54301
+RNApolis-0.3.17.dist-info/WHEEL,sha256=mguMlWGMX-VHnMpKOjjQidIo1ssRlCFu4a4mBpz1s2M,91
+RNApolis-0.3.17.dist-info/entry_points.txt,sha256=foN2Pn5e-OzEz0fFmNoX6PnFSZFQntOlY8LbognP5F0,308
+RNApolis-0.3.17.dist-info/top_level.txt,sha256=LcO18koxZcWoJ21KDRRRo_tyIbmXL5z61dPitZpy8yc,9
+RNApolis-0.3.17.dist-info/RECORD,,

{RNApolis-0.3.15.dist-info → RNApolis-0.3.17.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: bdist_wheel (0.43.0)
+Generator: setuptools (70.1.1)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

rnapolis/metareader.py

@@ -3,6 +3,7 @@ import argparse
 from typing import IO, Dict, List
 
 import orjson
+import pandas as pd
 from mmcif.io.IoAdapterPy import IoAdapterPy
 from mmcif.io.PdbxReader import DataContainer
 
@@ -46,6 +47,10 @@ def main():
         help="read the mmCIF file and list categories available inside",
         action="store_true",
     )
+    parser.add_argument(
+        "--csv-directory",
+        help="directory where to output CSV per each category",
+    )
     args = parser.parse_args()
 
     file = handle_input_file(args.path)
@@ -57,6 +62,12 @@ def main():
         result = read_metadata(file, args.category)
         print(orjson.dumps(result).decode("utf-8"))
 
+        if args.csv_directory:
+            for category in result:
+                with open(f"{args.csv_directory}/{category}.csv", "w") as f:
+                    df = pd.DataFrame(result[category])
+                    df.to_csv(f, index=False)
+
 
 if __name__ == "__main__":
     main()

rnapolis/parser.py

@@ -1,3 +1,4 @@
+import logging
 from typing import IO, Dict, List, Optional, Tuple, Union
 
 from mmcif.io.IoAdapterPy import IoAdapterPy
@@ -5,6 +6,9 @@ from mmcif.io.IoAdapterPy import IoAdapterPy
 from rnapolis.common import ResidueAuth, ResidueLabel
 from rnapolis.tertiary import BASE_ATOMS, Atom, Residue3D, Structure3D
 
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
 
 def read_3d_structure(
     cif_or_pdb: IO[str], model: Optional[int] = None, nucleic_acid_only: bool = False
@@ -104,6 +108,14 @@ def parse_cif(
                         auth_residue_name,
                     )
 
+                if label is None and auth is None:
+                    # this should not happen in a valid mmCIF file
+                    # skipping the line
+                    logger.debug(
+                        f"Cannot parse an atom line without chain name, residue number, and residue name: {row}"
+                    )
+                    continue
+
                 model = int(row_dict.get("pdbx_PDB_model_num", "1"))
                 atom_name = row_dict["label_atom_id"]
                 x = float(row_dict["Cartn_x"])
@@ -200,7 +212,7 @@ def parse_pdb(
             if alternate_location != " ":
                 continue
             atom_name = line[12:16].strip()
-            residue_name = line[18:20].strip()
+            residue_name = line[17:20].strip()
             chain_identifier = line[21]
             residue_number = int(line[22:26].strip())
             insertion_code = line[26] if line[26] != " " else None