Cast line_idx to int64 in empty PDB record frames

biopandas/pdb/pandas_pdb.py

@@ -564,6 +564,7 @@ def _construct_df(pdb_lines):
                 except ValueError:
                     # expect ValueError if float/int columns are empty strings
                     df[c["id"]] = pd.Series(np.nan, index=df.index)
+            df["line_idx"] = df["line_idx"].astype(np.int64)
 
             dfs[r[0]] = df
 

docs/CHANGELOG.md

@@ -13,6 +13,8 @@ The CHANGELOG for the current development version is available at
 
 - Fix: `PandasPdb.to_pdb()` no longer fails with a `TypeError` on structures with `ANISOU` records after `label_models()`. The `model_id` column is now skipped silently, like in `to_pdb_stream()`, and other unexpected `ANISOU` columns are skipped with a warning, as for `ATOM` and `HETATM`. Addresses issue [#124](https://github.com/BioPandas/biopandas/issues/124).
 
+- Fix: `line_idx` is now `int64` in empty record DataFrames (e.g. `ANISOU` for a file without anisotropic records) instead of `object`, matching non-empty frames. Addresses issue [#64](https://github.com/BioPandas/biopandas/issues/64).
+
 ### 0.5.2 (24/08/2026)
 
 - Fix: improves robustness of `amino3to1` residue identification by handling `NaN` insertion codes (NumPy 2.4+ compatibility) and including chain IDs and separators to avoid collisions. PR [#157](https://github.com/BioPandas/biopandas/pull/157)

tests/pdb/test_read_pdb.py

@@ -271,6 +271,14 @@ def test_read_pdb_from_list():
         assert ppdb.pdb_path == ""
 
 
+def test_line_idx_dtype_for_empty_records():
+    ppdb = PandasPdb()
+    ppdb.read_pdb(TESTDATA_FILENAME)
+    assert ppdb.df["ANISOU"].empty
+    for record in ("ATOM", "HETATM", "ANISOU", "OTHERS"):
+        assert ppdb.df[record]["line_idx"].dtype == np.int64
+
+
 def test_anisou_input_handling():
     """Test public read_pdb"""
     ppdb = PandasPdb()