scrise 1.2.0__tar.gz

scrise-1.2.0/PKG-INFO

@@ -0,0 +1,100 @@
+Metadata-Version: 2.4
+Name: scrise
+Version: 1.2.0
+Summary: Single cell analysis across conditions using PARAFAC2.
+Author: Andrew Ramirez, Aaron Meyer
+Author-email: Aaron Meyer <git@asmlab.org>
+License-Expression: MIT
+Requires-Dist: numpy>=2.2
+Requires-Dist: scipy>=1.16
+Requires-Dist: scikit-learn>=1.6
+Requires-Dist: pandas>=3.0.0
+Requires-Dist: pyarrow>=19.0
+Requires-Dist: tensorly>=0.9.0
+Requires-Dist: parafac2>=1.6.0
+Requires-Dist: anndata>=0.13
+Requires-Dist: pacmap>=0.9
+Requires-Dist: tqdm>=4.66.1
+Requires-Dist: vcsc
+Requires-Dist: parafac2[gpu] ; extra == 'gpu'
+Requires-Python: >=3.13
+Provides-Extra: gpu
+Description-Content-Type: text/markdown
+
+# RISE - Reduction and Insight in Single-cell Exploration
+
+RISE (Reduction and Insight in Single-cell Exploration) is an unsupervised, tensor-based computational method designed for the integrative analysis of single-cell RNA sequencing (scRNA-seq) data across multiple experimental conditions, such as drug treatments, patient cohorts, or time points. Built upon the PARAFAC2 tensor decomposition framework, RISE preserves the inherent three-dimensional structure of multi-condition single-cell data—conditions × cells × genes—instead of flattening it into a conventional two-dimensional matrix. This allows RISE to decompose variation into distinct, interpretable patterns associated with experimental conditions, individual cells, and genes, providing a more nuanced and biologically meaningful analysis.
+
+RISE does not require prior cell-type labels or clustering, reducing bias and enabling discovery of novel cell states, while also separating technical, biological, and condition-driven variation without batch correction that may erase meaningful signals. Its high resolution enables the identification of cell populations and condition-specific subpopulations missed by pseudobulk or clustering-based approaches, and each resulting component is directly linked to specific conditions, genes, and cells, making the results biologically tractable.
+
+- **Read the documentation** at [RISE Documentation](https://meyer-lab.github.io/RISE/).
+- RISE uses the [AnnData](https://anndata.readthedocs.io/) format for handling single-cell data matrices.
+
+## Installation
+
+> **Note:** The `RISE` package was renamed to `scrise` on PyPI (the import name changed from `RISE` to `scrise`). The GitHub repository name is unchanged. If you have `RISE` pinned in a `requirements.txt` or install script, update it to `scrise` as shown below.
+
+To add `scrise` to your Python environment, install it from PyPI:
+
+```bash
+pip install scrise
+```
+
+For GPU acceleration support (propagated to `parafac2[gpu]`):
+
+```bash
+pip install "scrise[gpu]"
+```
+
+Or add the following line to your `requirements.txt`:
+
+```
+scrise
+```
+
+or with GPU support:
+
+```
+scrise[gpu]
+```
+
+
+## Quick Start
+
+RISE works with preprocessed AnnData objects containing single-cell RNA-seq data:
+
+```python
+from scrise.factorization import pf2
+
+# Perform PARAFAC2 tensor decomposition
+X = pf2(X=adata, rank=20, doEmbedding=True, random_state=42)
+
+# Results are stored in the AnnData object:
+# - X.uns["Pf2_weights"]: Component weights
+# - X.uns["Pf2_A"]: Condition factors
+# - X.uns["Pf2_B"]: Eigen-state factors
+# - X.varm["Pf2_C"]: Gene factors
+# - X.obsm["projections"]: Cell projections
+# - X.obsm["weighted_projections"]: Weighted cell projections
+```
+
+See the [tutorial](https://meyer-lab.github.io/RISE/tutorial.html) for a complete workflow including preprocessing, rank selection, visualization, and interpretation.
+
+## Key Features
+
+- **Tensor-based decomposition**: Preserves the 3D structure of multi-condition scRNA-seq data
+- **Unsupervised analysis**: No prior cell-type labels or clustering required
+- **High resolution**: Identifies cell populations and condition-specific subpopulations
+- **Interpretable results**: Components directly linked to conditions, cells, and genes
+- **Integrated workflow**: Built-in preprocessing, visualization, and interpretation tools
+- **Principled rank selection**: Bi-cross-validation (`scrise.rank_selection`) for choosing the number of components by evaluating a set of candidate ranks
+
+## Citation
+
+If you use RISE in your work, please cite the RISE publication as follows:
+
+**Integrative, high-resolution analysis of single-cell gene expression across experimental conditions with PARAFAC2-RISE**
+
+Andrew Ramirez, [...], Aaron Meyer
+
+*Cell Systems*, 2025. DOI: [10.1016/j.cels.2025.101294](https://doi.org/10.1016/j.cels.2025.101294)

scrise-1.2.0/README.md

@@ -0,0 +1,77 @@
+# RISE - Reduction and Insight in Single-cell Exploration
+
+RISE (Reduction and Insight in Single-cell Exploration) is an unsupervised, tensor-based computational method designed for the integrative analysis of single-cell RNA sequencing (scRNA-seq) data across multiple experimental conditions, such as drug treatments, patient cohorts, or time points. Built upon the PARAFAC2 tensor decomposition framework, RISE preserves the inherent three-dimensional structure of multi-condition single-cell data—conditions × cells × genes—instead of flattening it into a conventional two-dimensional matrix. This allows RISE to decompose variation into distinct, interpretable patterns associated with experimental conditions, individual cells, and genes, providing a more nuanced and biologically meaningful analysis.
+
+RISE does not require prior cell-type labels or clustering, reducing bias and enabling discovery of novel cell states, while also separating technical, biological, and condition-driven variation without batch correction that may erase meaningful signals. Its high resolution enables the identification of cell populations and condition-specific subpopulations missed by pseudobulk or clustering-based approaches, and each resulting component is directly linked to specific conditions, genes, and cells, making the results biologically tractable.
+
+- **Read the documentation** at [RISE Documentation](https://meyer-lab.github.io/RISE/).
+- RISE uses the [AnnData](https://anndata.readthedocs.io/) format for handling single-cell data matrices.
+
+## Installation
+
+> **Note:** The `RISE` package was renamed to `scrise` on PyPI (the import name changed from `RISE` to `scrise`). The GitHub repository name is unchanged. If you have `RISE` pinned in a `requirements.txt` or install script, update it to `scrise` as shown below.
+
+To add `scrise` to your Python environment, install it from PyPI:
+
+```bash
+pip install scrise
+```
+
+For GPU acceleration support (propagated to `parafac2[gpu]`):
+
+```bash
+pip install "scrise[gpu]"
+```
+
+Or add the following line to your `requirements.txt`:
+
+```
+scrise
+```
+
+or with GPU support:
+
+```
+scrise[gpu]
+```
+
+
+## Quick Start
+
+RISE works with preprocessed AnnData objects containing single-cell RNA-seq data:
+
+```python
+from scrise.factorization import pf2
+
+# Perform PARAFAC2 tensor decomposition
+X = pf2(X=adata, rank=20, doEmbedding=True, random_state=42)
+
+# Results are stored in the AnnData object:
+# - X.uns["Pf2_weights"]: Component weights
+# - X.uns["Pf2_A"]: Condition factors
+# - X.uns["Pf2_B"]: Eigen-state factors
+# - X.varm["Pf2_C"]: Gene factors
+# - X.obsm["projections"]: Cell projections
+# - X.obsm["weighted_projections"]: Weighted cell projections
+```
+
+See the [tutorial](https://meyer-lab.github.io/RISE/tutorial.html) for a complete workflow including preprocessing, rank selection, visualization, and interpretation.
+
+## Key Features
+
+- **Tensor-based decomposition**: Preserves the 3D structure of multi-condition scRNA-seq data
+- **Unsupervised analysis**: No prior cell-type labels or clustering required
+- **High resolution**: Identifies cell populations and condition-specific subpopulations
+- **Interpretable results**: Components directly linked to conditions, cells, and genes
+- **Integrated workflow**: Built-in preprocessing, visualization, and interpretation tools
+- **Principled rank selection**: Bi-cross-validation (`scrise.rank_selection`) for choosing the number of components by evaluating a set of candidate ranks
+
+## Citation
+
+If you use RISE in your work, please cite the RISE publication as follows:
+
+**Integrative, high-resolution analysis of single-cell gene expression across experimental conditions with PARAFAC2-RISE**
+
+Andrew Ramirez, [...], Aaron Meyer
+
+*Cell Systems*, 2025. DOI: [10.1016/j.cels.2025.101294](https://doi.org/10.1016/j.cels.2025.101294)

scrise-1.2.0/pyproject.toml

@@ -0,0 +1,92 @@
+[project]
+name = "scrise"
+version = "1.2.0"
+description = "Single cell analysis across conditions using PARAFAC2."
+license = "MIT"
+requires-python = ">= 3.13"
+dependencies = [
+    "numpy>=2.2",
+    "scipy>=1.16",
+    "scikit-learn>=1.6",
+    "pandas>=3.0.0",
+    "pyarrow>=19.0",
+    "tensorly>=0.9.0",
+    "parafac2>=1.6.0",
+    "anndata>=0.13",
+    "pacmap>=0.9",
+    "tqdm>=4.66.1",
+    "vcsc",
+]
+readme = "README.md"
+
+[[project.authors]]
+name = "Andrew Ramirez"
+
+[[project.authors]]
+name = "Aaron Meyer"
+email = "git@asmlab.org"
+
+[project.optional-dependencies]
+gpu = ["parafac2[gpu]"]
+
+[project.scripts]
+fbuild = "analysis.figures.common:genFigure"
+
+[build-system]
+requires = ["uv_build>=0.12.0,<0.13"]
+build-backend = "uv_build"
+
+[tool.uv.build-backend]
+module-name = ["scrise"]
+module-root = ""
+
+[tool.uv.sources.vcsc]
+git = "https://github.com/meyer-lab/anndata-VCSC.git"
+
+[tool.ruff.lint]
+select = [
+    "E",
+    "F",
+    "I",
+    "UP",
+    "B",
+]
+ignore = ["E501"]
+
+[tool.pytest.ini_options]
+filterwarnings = [
+    "ignore::anndata.OldFormatWarning",
+    "ignore::FutureWarning",
+    "ignore::UserWarning",
+    "ignore::DeprecationWarning",
+    "ignore::PendingDeprecationWarning:seaborn",
+]
+
+[dependency-groups]
+analysis = [
+    "seaborn>=0.13.2",
+    "datashader>=0.19",
+    "scanpy>=1.12",
+    "tlviz>=0.1.1",
+    "doubletdetection>=4.3",
+    "statsmodels>=0.14.1",
+    "hdf5plugin>=7.0.0",
+]
+benchmarking = [
+    "scanorama>=1.7.4",
+    "harmonypy>=0.0.10",
+    "scib>=1.1.5",
+]
+docs = [
+    { include-group = "analysis" },
+    "mkdocs>=1.6",
+    "mkdocs-material>=9.5",
+    "mkdocstrings[python]>=0.28",
+]
+dev = [
+    { include-group = "analysis" },
+    "pytest>=9.0",
+    "pytest-cov>=7.0",
+    "ty",
+    "ruff>=0.16",
+]

scrise-1.2.0/pyproject.toml.orig

@@ -0,0 +1,86 @@
+[project]
+name = "scrise"
+version = "1.2.0"
+description = "Single cell analysis across conditions using PARAFAC2."
+authors = [{name = "Andrew Ramirez"}, {name = "Aaron Meyer", email = "git@asmlab.org" }]
+license = "MIT"
+requires-python = ">= 3.13"
+
+dependencies = [
+    "numpy>=2.2",
+    "scipy>=1.16",
+    "scikit-learn>=1.6",
+    "pandas>=3.0.0",
+    "pyarrow>=19.0",
+    "tensorly>=0.9.0",
+    "parafac2>=1.6.0",
+    "anndata>=0.13",
+    "pacmap>=0.9",
+    "tqdm>=4.66.1",
+    "vcsc",
+]
+
+readme = "README.md"
+
+[project.optional-dependencies]
+gpu = [
+    "parafac2[gpu]",
+]
+
+
+[project.scripts]
+fbuild = "analysis.figures.common:genFigure"
+
+[build-system]
+requires = ["uv_build>=0.12.0,<0.13"]
+build-backend = "uv_build"
+
+[tool.uv.build-backend]
+module-name = ["scrise"]
+module-root = ""
+
+[tool.uv.sources]
+vcsc = { git = "https://github.com/meyer-lab/anndata-VCSC.git" }
+
+
+[dependency-groups]
+analysis = [
+    "seaborn>=0.13.2",
+    "datashader>=0.19",
+    "scanpy>=1.12",
+    "tlviz>=0.1.1",
+    "doubletdetection>=4.3",
+    "statsmodels>=0.14.1",
+    "hdf5plugin>=7.0.0",
+]
+benchmarking = [
+    "scanorama>=1.7.4",
+    "harmonypy>=0.0.10",
+    "scib>=1.1.5",
+]
+docs = [
+    { include-group = "analysis" },
+    "mkdocs>=1.6",
+    "mkdocs-material>=9.5",
+    "mkdocstrings[python]>=0.28",
+]
+dev = [
+    { include-group = "analysis" },
+    "pytest>=9.0",
+    "pytest-cov>=7.0",
+    "ty",
+    "ruff>=0.16",
+]
+
+[tool.ruff.lint]
+select = ["E", "F", "I", "UP", "B"]
+ignore = ["E501"]
+
+[tool.pytest.ini_options]
+filterwarnings = [
+    "ignore::anndata.OldFormatWarning",
+    "ignore::FutureWarning",
+    "ignore::UserWarning",
+    "ignore::DeprecationWarning",
+    "ignore::PendingDeprecationWarning:seaborn",
+]

scrise-1.2.0/scrise/__init__.py

@@ -0,0 +1,36 @@
+from .annotation_alignment import (
+    CellTypeAlignmentResults,
+    ComponentAlignmentResult,
+    cell_type_alignment,
+    compute_tau,
+    score_cell_type_alignment,
+)
+from .factorization import (
+    canonical_component_signs,
+    correct_conditions,
+    export_factors,
+    load_factors,
+    match_components_across_ranks,
+    order_components_by_energy,
+    pf2,
+    rise_pca_r2x,
+)
+from .opq import OPQQuantizer, find_optimal_opq
+
+__all__ = [
+    "CellTypeAlignmentResults",
+    "ComponentAlignmentResult",
+    "OPQQuantizer",
+    "canonical_component_signs",
+    "cell_type_alignment",
+    "compute_tau",
+    "correct_conditions",
+    "export_factors",
+    "find_optimal_opq",
+    "load_factors",
+    "match_components_across_ranks",
+    "order_components_by_energy",
+    "pf2",
+    "rise_pca_r2x",
+    "score_cell_type_alignment",
+]