bumped version to 0.4.0

src/tfscreen/__version__.py

@@ -5,5 +5,5 @@
 Version string. 
 """
 
-VERSION = (0, 3, 1)
+VERSION = (0, 4, 0)
 __version__ = '.'.join(map(str, VERSION))

src/tfscreen/genetics/combine_mutation_effects.py

@@ -68,7 +68,8 @@ def combine_mutation_effects(
         g_lookup.loc["wt", :] = np.nan
 
     # Convert from wide (genotypes) to long (single mutations) format
-    stacked_muts = g_lookup.stack()
+    # dropna() required because pandas 3.x stack() no longer drops NaN by default
+    stacked_muts = g_lookup.stack().dropna()
     if stacked_muts.empty: # Case where only 'wt' or no multi-mutants exist
          return single_mutant_effects.iloc[0:0].reindex(g_lookup.index, fill_value=0)
 

src/tfscreen/simulate/selection_experiment.py

@@ -720,7 +720,8 @@ def _sim_growth(
     all_kt = genotype_vs_kt[transformants]
 
     # Expand the 2D mask to 3D to match the shape of all_kt.
-    expanded_mask = np.broadcast_to(trans_mask[:, :, np.newaxis], all_kt.shape)
+    # .copy() required: numpy 2.x rejects read-only broadcast arrays as ma masks
+    expanded_mask = np.broadcast_to(trans_mask[:, :, np.newaxis], all_kt.shape).copy()
     masked_kt = ma.array(all_kt, mask=expanded_mask)
 
     # Calculate the kt for each cell by combining effects of plasmids
@@ -1054,8 +1055,9 @@ def _simulate_library_group(
     # stack, these will be removed. 
     genotype_vs_kt_pivot = sparse_pivot.reindex(ordered_genotypes, fill_value=0)
 
-    # Create a 2D array of (genotype,conditions) holding kt. 
-    genotype_vs_kt = genotype_vs_kt_pivot.to_numpy()
+    # Create a 2D array of (genotype,conditions) holding kt.
+    # .copy() required: pandas 3.x CoW returns a read-only view from to_numpy()
+    genotype_vs_kt = genotype_vs_kt_pivot.to_numpy().copy()
 
     # Add per-tube environmental noise: one delta_k per condition, shared across
     # all genotypes in that tube.  tube_noise_sigma is in growth-rate units

tests/tfscreen/tfmodel/components/transformation/test__congression.py

@@ -38,7 +38,7 @@ def test_empirical_cdf():
     assert cdf.shape == (5,)
     assert cdf[0] == 0.16666667 # Interp 0.0 at 0.1 returns y[0]
     assert jnp.isclose(cdf[2], 0.5) # 0.5 is exactly in theta
-    assert cdf[4] == 0.8333333 # Interp 1.0 at 0.9 returns y[2]
+    assert jnp.isclose(cdf[4], 5/6, atol=1e-5) # Interp 1.0 at 0.9 returns y[2]
 
 
 # -------------------------------------------------------------------------

tests/tfscreen/util/io/test_read_dataframe.py

@@ -211,5 +211,5 @@ def test_genotype_as_index_not_categorized():
     df = pd.DataFrame({"genotype": ["wt", "A1T"], "value": [1, 2]})
     result = read_dataframe(df, index_column="genotype")
     assert result.index.name == "genotype"
-    # Index should be plain strings, not categorical
-    assert result.index.dtype == object
+    # Index should be plain strings, not categorical (dtype varies: object or StringDtype)
+    assert not isinstance(result.index.dtype, pd.CategoricalDtype)