rectanglepy 1.2.1__tar.gz → 1.3.0__tar.gz

{rectanglepy-1.2.1 → rectanglepy-1.3.0}/.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 1.2.1
+current_version = 1.3.0
 tag = True
 commit = True
 

{rectanglepy-1.2.1 → rectanglepy-1.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rectanglepy
-Version: 1.2.1
+Version: 1.3.0
 Summary: Hierarchical deconvolution of bulk  transcriptomics
 Project-URL: Documentation, https://rectanglepy.readthedocs.io/
 Project-URL: Source, https://github.com/ComputationalBiomedicineGroup/Rectangle
@@ -690,7 +690,6 @@ Requires-Dist: osqp>=1.0.5
 Requires-Dist: pandas<3.0.0,>=2.0.0
 Requires-Dist: pydeseq2==0.4.11
 Requires-Dist: scipy==1.13.0
-Requires-Dist: setuptools>=80.0.0
 Requires-Dist: statsmodels>=0.14.1
 Provides-Extra: dev
 Requires-Dist: bump2version; extra == 'dev'

{rectanglepy-1.2.1 → rectanglepy-1.3.0}/pyproject.toml

@@ -4,7 +4,7 @@ requires = ["hatchling"]
 
 [project]
 name = "rectanglepy"
-version = "1.2.1"
+version = "1.3.0"
 description = "Hierarchical deconvolution of bulk  transcriptomics"
 readme = "README.md"
 requires-python = ">=3.10"
@@ -27,7 +27,6 @@ dependencies = [
     "numpy>=1.0.0,<2.0.0",
     "anndata>=0.8.0,<0.10.9",
     "statsmodels>=0.14.1",
-    "setuptools>=80.0.0",
 
 ]
 

{rectanglepy-1.2.1 → rectanglepy-1.3.0}/src/rectanglepy/rectangle.py

@@ -1,8 +1,9 @@
+from importlib.resources import files
+
 import pandas as pd
 from anndata import AnnData
 from loguru import logger
 from pandas import DataFrame
-from pkg_resources import resource_stream
 
 from .pp import RectangleSignatureResult, build_rectangle_signatures
 from .tl import deconvolution
@@ -93,13 +94,15 @@ def load_tutorial_data() -> tuple[pd.DataFrame, pd.DataFrame, pd.DataFrame]:
     -------
     The single-cell count data, annotations, and bulk data.
     """
-    with resource_stream(__name__, "data/hao1_annotations_small.zip") as annotations_file:
+    data_dir = files(__package__) / "data"
+
+    with (data_dir / "hao1_annotations_small.zip").open("rb") as annotations_file:
         annotations = pd.read_csv(annotations_file, index_col=0, compression="zip")["0"]
 
-    with resource_stream(__name__, "data/hao1_counts_small.zip") as counts_file:
+    with (data_dir / "hao1_counts_small.zip").open("rb") as counts_file:
         sc_counts = pd.read_csv(counts_file, index_col=0, compression="zip").astype("int")
 
-    with resource_stream(__name__, "data/small_fino_bulks.zip") as bulks_file:
+    with (data_dir / "small_fino_bulks.zip").open("rb") as bulks_file:
         bulks = pd.read_csv(bulks_file, index_col=0, compression="zip")
 
     return sc_counts.T, annotations, bulks.T

{rectanglepy-1.2.1 → rectanglepy-1.3.0}/src/rectanglepy/tl/deconvolution.py

@@ -121,13 +121,18 @@ def solve_qp(
         P = P / scale
         q = q / scale
 
+    # OSQP can return noticeably different solutions from active-set QP when P is singular/ill-conditioned.
+    # Add tiny ridge to make the problem strictly convex and closer to quadprog behavior.
+    ridge = 1e-8
+    P = ((P + P.T) / 2.0) + ridge * np.eye(n_vars, dtype=np.float64)
+
     # OSQP uses l <= A x <= u
     A = C.T  # (n_constraints, n_vars)
     l = b
     u = np.full_like(l, np.inf, dtype=np.float64)
 
     # Sparse matrices (required/expected)
-    P_sp = sp.csc_matrix((P + P.T) / 2.0)  # enforce symmetry
+    P_sp = sp.csc_matrix(P)
     A_sp = sp.csc_matrix(A)
 
     solver = osqp.OSQP()
@@ -138,9 +143,12 @@ def solve_qp(
         l=l,
         u=u,
         verbose=False,
-        eps_abs=1e-5,
-        eps_rel=1e-5,
-        max_iter=10000,
+        eps_abs=1e-8,
+        eps_rel=1e-8,
+        max_iter=50000,
+        polish=True,
+        warm_start=False,
+        scaled_termination=False,
     )
 
     res = solver.solve()