stgp 0.1.0__tar.gz

stgp-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Baichen Yu and contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

stgp-0.1.0/PKG-INFO

@@ -0,0 +1,53 @@
+Metadata-Version: 2.4
+Name: stgp
+Version: 0.1.0
+Summary: Spatiotemporal gene program discovery with Gaussian process priors
+Author-email: Baichen Yu <mabyu@ust.hk>
+Maintainer-email: Baichen Yu <mabyu@ust.hk>
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/YangLabHKUST/stGP
+Project-URL: Repository, https://github.com/YangLabHKUST/stGP
+Keywords: spatial transcriptomics,gene programs,gaussian process,matrix factorization
+Classifier: Development Status :: 3 - Alpha
+Classifier: Intended Audience :: Science/Research
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: numpy<2.0,>=1.22
+Requires-Dist: scipy>=1.10
+Dynamic: license-file
+
+# stGP: Characterizing dynamic tissue architectures by identifying cell-type-specific spatiotemporal gene programs
+
+## Introduction
+
+stGP is a statistical framework for identifying interpretable cell-type-specific spatiotemporal gene programs (stGPs) from multi-sample spatiotemporal transcriptomic data measured across biological time by deciphering temporal trajectory and dynamic spatial patterns.
+
+stGP's effectiveness relies on our innovations in the integration of Gaussian process priors and interpretable matrix factorization:
+
+- stGP represents gene expression within each cell type as a small set of latent programs with non-negative gene loadings, making each program interpretable as a weighted gene set shared across samples. Variance components quantify the relative contributions of time and space to each program.
+- stGP decomposes per-cell program activity into a sample-level temporal component that captures coordinated responses over biological time (e.g., age or stage), and a within-section spatial component that characterizes dynamic program deployment across tissue coordinates without requiring cross-section registration.
+- For multi-program inference, stGP adopts a blockwise backfitting scheme that sequentially extracts rank-1 components from residuals, with automatic model selection to determine the number of programs.
+
+
+## Reference
+
+If you find `stGP` useful for your work, please cite:
+> Characterizing dynamic tissue architectures by identifying cell-type-specific spatiotemporal gene programs with stGP.
+> Baichen Yu, Ziyue Tan, Xiaomeng Wan, Hansheng Wang, and Can Yang.
+> Working paper, 2026.
+
+## Development
+
+The software is developed and maintained by [Baichen Yu](mailto:mabyu@ust.hk).
+
+## Contact
+
+Please feel free to contact [Baichen Yu](mailto:mabyu@ust.hk) or [Prof. Can Yang](mailto:macyang@ust.hk) if any inquiries.

stgp-0.1.0/README.md

@@ -0,0 +1,82 @@
+# Characterizing dynamic tissue architectures by identifying cell-type-specific spatiotemporal gene programs with stGP
+
+This repository contains the code for **stGP** (spatiotemporal Gene Programs for spatial transcriptomics by Gaussian Process), a statistical framework for discovering spatiotemporally variable gene programs from aging spatial transcriptomic data.
+
+## Introduction
+
+stGP is a statistical framework for identifying interpretable cell-type-specific spatiotemporal gene programs (stGPs) from multi-sample spatiotemporal transcriptomic data measured across biological time.
+
+stGP's effectiveness relies on our innovations in the integration of Gaussian process priors and interpretable matrix factorization:
+
+- stGP represents gene expression within each cell type as a small set of latent programs with non-negative gene loadings constrained to the simplex, making each program interpretable as a weighted gene set shared across samples.
+- stGP decomposes per-cell program activity into a sample-level temporal component that captures coordinated responses over biological time (e.g., age or stage), and a within-section spatial component that characterizes local program deployment across tissue coordinates—without requiring cross-section registration.
+- Gaussian process priors smooth temporal trajectories along the biological-time covariate and spatial embeddings within each tissue section, while variance components quantify the relative contributions of time and space to each program.
+- For multi-program inference, stGP adopts a blockwise backfitting scheme that sequentially extracts rank-1 components from residuals, with automatic model selection to determine the number of programs.
+
+<figure>
+
+<img src="FigureReproducing/Fig1_overview.png" style="width:85.0%"
+alt="Overview" />
+</figure>
+
+
+## Installation
+
+```bash
+# Clone or copy the repository
+cd stGP
+conda env create -f stGP.yml
+conda activate stGP
+```
+
+### Popari environment (separate)
+
+Popari requires an isolated conda environment due to PyTorch/SIGFPE conflicts.
+
+```bash
+git clone https://github.com/alam-shahul/popari.git
+cd popari
+conda create -n Popari python=3.12
+conda activate Popari
+pip install uv
+uv pip install ".[mlflow,simulation]"
+pip install torch==2.6.0 --index-url https://download.pytorch.org/whl/cu126
+cd ..
+```
+
+## Quick start: simulation
+
+```bash
+cd Simulation
+
+# Count-regime multi-program recovery
+python simu_gaussian.py --stage all --methods auto --reps 50
+```
+
+## Quick start: real data
+
+```bash
+cd RealData_MouseBrainMERFISH
+
+# Run steps individually for one cell type
+python3 01_preprocess_qc.py --output data/qc/aging_coronal_qc.h5ad
+python3 02_run_stgp.py --celltypes Microglia
+python3 04_make_figures.py --celltype Microglia
+
+# Baselines and downstream analyses are documented in RealData_MouseBrainMERFISH/README.md
+```
+
+## Reference
+
+If you find `stGP` or any of the source code in this repository useful for your work, please cite:
+> Characterizing dynamic tissue architectures by identifying cell-type-specific spatiotemporal gene programs with stGP.
+> Baichen Yu, Ziyue Tan, Xiaomeng Wan, Hansheng Wang, and Can Yang.
+> Working paper, 2026.
+
+## Development
+
+The software is developed and maintained by [Baichen Yu](mailto:mabyu@ust.hk).
+
+## Contact
+
+Please feel free to contact [Baichen Yu](mailto:mabyu@ust.hk) or [Prof. Can Yang](mailto:macyang@ust.hk) if any inquiries.

stgp-0.1.0/pyproject.toml

@@ -0,0 +1,45 @@
+[build-system]
+requires = ["setuptools>=77", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "stgp"
+version = "0.1.0"
+description = "Spatiotemporal gene program discovery with Gaussian process priors"
+readme = "stgp/README.md"
+requires-python = ">=3.9"
+license = "MIT"
+authors = [
+    { name = "Baichen Yu", email = "mabyu@ust.hk" },
+]
+maintainers = [
+    { name = "Baichen Yu", email = "mabyu@ust.hk" },
+]
+dependencies = [
+    "numpy>=1.22,<2.0",
+    "scipy>=1.10",
+]
+keywords = [
+    "spatial transcriptomics",
+    "gene programs",
+    "gaussian process",
+    "matrix factorization",
+]
+classifiers = [
+    "Development Status :: 3 - Alpha",
+    "Intended Audience :: Science/Research",
+    "Operating System :: OS Independent",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.9",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Topic :: Scientific/Engineering :: Bio-Informatics",
+]
+
+[project.urls]
+Homepage = "https://github.com/YangLabHKUST/stGP"
+Repository = "https://github.com/YangLabHKUST/stGP"
+
+[tool.setuptools.packages.find]
+include = ["stgp*"]

stgp-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

stgp-0.1.0/stgp/README.md

@@ -0,0 +1,27 @@
+# stGP: Characterizing dynamic tissue architectures by identifying cell-type-specific spatiotemporal gene programs
+
+## Introduction
+
+stGP is a statistical framework for identifying interpretable cell-type-specific spatiotemporal gene programs (stGPs) from multi-sample spatiotemporal transcriptomic data measured across biological time by deciphering temporal trajectory and dynamic spatial patterns.
+
+stGP's effectiveness relies on our innovations in the integration of Gaussian process priors and interpretable matrix factorization:
+
+- stGP represents gene expression within each cell type as a small set of latent programs with non-negative gene loadings, making each program interpretable as a weighted gene set shared across samples. Variance components quantify the relative contributions of time and space to each program.
+- stGP decomposes per-cell program activity into a sample-level temporal component that captures coordinated responses over biological time (e.g., age or stage), and a within-section spatial component that characterizes dynamic program deployment across tissue coordinates without requiring cross-section registration.
+- For multi-program inference, stGP adopts a blockwise backfitting scheme that sequentially extracts rank-1 components from residuals, with automatic model selection to determine the number of programs.
+
+
+## Reference
+
+If you find `stGP` useful for your work, please cite:
+> Characterizing dynamic tissue architectures by identifying cell-type-specific spatiotemporal gene programs with stGP.
+> Baichen Yu, Ziyue Tan, Xiaomeng Wan, Hansheng Wang, and Can Yang.
+> Working paper, 2026.
+
+## Development
+
+The software is developed and maintained by [Baichen Yu](mailto:mabyu@ust.hk).
+
+## Contact
+
+Please feel free to contact [Baichen Yu](mailto:mabyu@ust.hk) or [Prof. Can Yang](mailto:macyang@ust.hk) if any inquiries.

stgp-0.1.0/stgp/__init__.py

@@ -0,0 +1,48 @@
+from stgp.estimation import (
+    fit_rank1,
+    fit_pfactor,
+    fit_pfactor_auto,
+    project_simplex,
+    project_simplex_topk,
+    recover_low_rank_signal,
+    align_programs_and_mse,
+)
+
+from stgp.kernels import (
+    build_K_age,
+    build_K_spa,
+    build_K_spa_list,
+    build_K_spa_list_from_stacked,
+    rbf_kernel_1d,
+    ar1_kernel_1d,
+)
+
+from stgp.preprocessing import (
+    log1p_normalize,
+    log1p_norm_centered_list,
+    demean_genes,
+    library_normalize,
+)
+
+__version__ = "0.1.0"
+
+__all__ = [
+    "__version__",
+    "fit_rank1",
+    "fit_pfactor",
+    "fit_pfactor_auto",
+    "project_simplex",
+    "project_simplex_topk",
+    "recover_low_rank_signal",
+    "align_programs_and_mse",
+    "build_K_age",
+    "build_K_spa",
+    "build_K_spa_list",
+    "build_K_spa_list_from_stacked",
+    "rbf_kernel_1d",
+    "ar1_kernel_1d",
+    "log1p_normalize",
+    "log1p_norm_centered_list",
+    "demean_genes",
+    "library_normalize",
+]