bsvae 0.4.3__tar.gz → 0.4.4__tar.gz

{bsvae-0.4.3 → bsvae-0.4.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bsvae
-Version: 0.4.3
+Version: 0.4.4
 Summary: Gaussian Mixture VAE for multi-modal biological module discovery in omics data.
 License: GPL-3.0-only
 License-File: LICENSE

{bsvae-0.4.3 → bsvae-0.4.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "bsvae"
-version = "0.4.3"  # dynamically resolved by poetry-dynamic-versioning
+version = "0.4.4"  # dynamically resolved by poetry-dynamic-versioning
 description = "Gaussian Mixture VAE for multi-modal biological module discovery in omics data."
 readme = "README.md"
 authors = ["Kynon J Benjamin <kj.benjamin90@gmail.com>"]

{bsvae-0.4.3 → bsvae-0.4.4}/src/bsvae/networks/targets.py

@@ -676,7 +676,10 @@ def combine_target_tests(
         null_like = zmat[(zmat.max(axis=1) < sps.norm.isf(0.05)) | zmat.isna().any(axis=1)]
         src = null_like if len(null_like) >= 20 else zmat
         rho = src.corr(min_periods=5).fillna(0.0).clip(lower=0.0)
-        np.fill_diagonal(rho.values, 1.0)
+        # write the unit diagonal into a copy: under pandas copy-on-write (the pandas 3 default) .values is read-only
+        r = rho.to_numpy(dtype=float, copy=True)
+        np.fill_diagonal(r, 1.0)
+        rho = pd.DataFrame(r, index=rho.index, columns=rho.columns)
     rows = []
     for tx, g in long.groupby(level=0):
         p = np.clip(g["p_value"].values.astype(float), 1e-300, 1.0)

{bsvae-0.4.3 → bsvae-0.4.4}/src/bsvae/utils/isoform.py

@@ -489,7 +489,8 @@ def isoform_specific_calls(
             E = _residualize_eigengenes(eigengenes, cols.index(int(gm))) if gm in cols else eigengenes
             kme[rows_idx] = compute_kme(resid.iloc[rows_idx], E, list(resid.index[rows_idx])).values
     else:
-        kme = compute_kme(resid, eigengenes, list(resid.index)).values
+        # a writable copy: the leave-one-out pass below updates kme in place (read-only view under copy-on-write)
+        kme = compute_kme(resid, eigengenes, list(resid.index)).to_numpy(dtype=np.float64, copy=True)
     mode = {True: "exact", False: "off"}.get(loo_eigengene, loo_eigengene)
     if mode not in ("exact", "approx", "off"):
         raise ValueError("loo_eigengene must be True/False, 'exact' or 'approx'")