scatrans 0.9.2.dev0__tar.gz → 0.9.4.2.dev0__tar.gz

scatrans-0.9.4.2.dev0/.github/workflows/ci.yml

@@ -0,0 +1,70 @@
+name: CI
+
+on:
+  push:
+    branches: [main, master]
+  pull_request:
+    branches: [main, master]
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ["3.9", "3.10", "3.11", "3.12"]
+        install-extra:
+          - ""
+          - "advanced"
+          - "pseudobulk"
+          - "gene_features"
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+          cache: "pip"
+
+      - name: Install package with extras
+        run: |
+          pip install --upgrade pip
+          EXTRA="${{ matrix.install-extra }}"
+          if [ -n "$EXTRA" ]; then
+            pip install -e ".[dev,$EXTRA]"
+          else
+            pip install -e ".[dev]"
+          fi
+
+      - name: Ruff lint
+        run: |
+          ruff check src/scatrans tests
+          ruff format --check src/scatrans tests
+
+      - name: Run fast tests
+        run: PYTHONPATH=src:. python -m pytest -m "not plot and not slow" -q
+
+      - name: Run slow tests (core / Python 3.11 only)
+        if: matrix.python-version == '3.11' && matrix.install-extra == ''
+        run: PYTHONPATH=src:. python -m pytest -m "slow and not plot" -q
+
+      - name: Run plot tests (core / Python 3.11 only)
+        if: matrix.python-version == '3.11' && matrix.install-extra == ''
+        run: PYTHONPATH=src:. python -m pytest -m plot -q
+
+      - name: Coverage report (core / Python 3.11 only)
+        if: matrix.python-version == '3.11' && matrix.install-extra == ''
+        run: |
+          PYTHONPATH=src:. python -m pytest -m "not plot" --cov=src/scatrans --cov-report=xml -q \
+            || PYTHONPATH=src:. python -m pytest -m "not plot" -q
+
+      - name: Upload coverage (optional)
+        if: matrix.python-version == '3.11' && matrix.install-extra == ''
+        uses: codecov/codecov-action@v4
+        with:
+          files: ./coverage.xml
+          fail_ci_if_error: false

scatrans-0.9.4.2.dev0/.gitignore

@@ -0,0 +1,177 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+scatrans.egg-info/
+src/scatrans.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+Pipfile.lock
+
+# poetry
+poetry.lock
+
+# pdm
+.pdm.toml
+.pdm-python
+.pdm-build/
+
+# PEP 582
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# Ruff
+.ruff_cache/
+
+# IDE / editors
+.idea/
+.vscode/
+*.swp
+*.swo
+*~
+
+# OS generated files
+.DS_Store
+.DS_Store?
+._*
+.Spotlight-V100
+.Trashes
+ehthumbs.db
+Thumbs.db
+
+# Project specific
+# Bundled large data is committed intentionally (gene features)
+# but do not add user-generated .parquet outputs here by mistake
+*.h5ad
+*.loom
+*.zarr
+# Keep source data/ but ignore generated outputs at root
+gene_features.parquet
+*_gene_features.parquet
+*_features.parquet
+# Bundled package data (must be tracked for CI / wheels)
+!src/scatrans/data/
+!src/scatrans/data/*.parquet
+# Temporary analysis outputs
+*.pdf
+*.png
+*.svg
+active_results/

{scatrans-0.9.2.dev0 → scatrans-0.9.4.2.dev0}/CHANGELOG.md

@@ -5,6 +5,58 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [Unreleased / Review 2026-06-27]
+### Added / Improved
+- Clarified and documented the `gamma_method="empirical_bayes"` implementation as **hierarchical (分层) gamma estimation** for the reference U/S ratio (README keeps the CN term; source now English-only).
+- Stronger emphasis on "always pass explicit target_group/reference_group".
+
+### Changed / Fixed (critical)
+- **Bug 1**: Eliminated dead code — `tl.py` now calls the canonical `run_permutation_test(...)` in `_permutation.py` instead of duplicating the Parallel loop. Removed ~duplicated logic and the maintenance trap. `valid_expr` is passed explicitly for consistent behavior.
+- **Bug 2**: Fixed double normalization (double log1p) in permutations. When `de_preprocess="normalize_log1p"` (or auto that applies), the value passed to permutation tasks is forced to "none" so that perm copies of the already-transformed adata are not re-normalized. Prevents systematically biased FDR.
+- **Scientific Error 1**: Fixed EB gamma: `sigma2` in `_apply_empirical_bayes_gamma` now correctly includes the `n_ref` factor: `1.0 / (n_ref * U_r + c) + ...`. This was causing n_r-fold over-estimate of observation noise and excessive shrinkage (especially bad for the small-ref case EB is meant to help). `n_ref` is computed from r_mask in the caller.
+- **Scientific Error 2**: Clarified `robust_median` docs in code/README: it is a heuristic variant of `heuristic_shrink` (different base_gamma estimator) and is **not** Bayesian/EB/hierarchical. Renamed descriptions to prevent user confusion.
+- **Design fixes**:
+  - Tightened Memento raw counts check from `>=` to exact `== n_vars` + `var_names` equality (prevents misaligned HVG/raw usage).
+  - Renamed shadowing local `ad = ...` in `active_score_simple` / `differential_expression_simple` (avoids hiding `import anndata as ad`).
+  - Made `min_cells_per_sample` private (`_min_cells_per_sample`) + doc note (was public but did nothing).
+  - Removed all Chinese comments ("分层") from source (tl.py, _velocity.py); retained in README only.
+  - Replaced risky bare `except TypeError` for PyDESeq2 design_factors/design compat with explicit `_pydeseq2_uses_design_factors()` using `importlib.metadata` version check.
+
+- Reduced anndata category storage log noise during internal DE/perm (narrow logger level bump).
+- All tests + targeted EB/perm/explicit-norm cases pass after fixes.
+
+### Additional fixes (2026-06-27 round 2)
+- **Scientific Error 3**: `generate_gene_features_from_gtf()` now computes **true exon union length per gene** (merge overlapping intervals) instead of naively summing all exons across all transcripts. Prevents ~N_transcript-fold overestimation of gene_length for multi-isoform genes (affects only users who generate their own tables; bundled .parquet files are unaffected).
+- **Scientific Error 4**: `logFC` is now normalized toward consistent **log2 scale** across backends inside `_run_de_wrapper`:
+  - wilcoxon, mixedlm, memento: divided by ln(2)
+  - t-test / PyDESeq2: left as native log2
+  - Updated docstring. Makes `logfc_cutoff` semantically comparable.
+- **Bug 3**: `_pseudobulk_with_layers` now uses a **per-run UUID-based private separator** (instead of fragile "||") to build internal keys. Completely eliminates split errors when sample names contain "||".
+- **Bug 4**: Removed redundant `import warnings as _w` inside `with warnings.catch_warnings()` in `_fit_huber_bias_correction`.
+- **Design 6**: `pval_cutoff` deprecation warning now fires **on every use** of the legacy name (previously only when != 0.05).
+- **Design 7**: `run_go(ontology="CC"/"MF")` now emits a clear INFO log explaining that only BP is bundled and the others require gseapy + network.
+- **Design 8**: Made the "max > 50" heuristic in `_prepare_log_normalized_expression` (mixed model prep) more robust: checks for negatives, uses lower threshold (20), better warning message.
+- **Design 9**: `valid_expr` is now **explicitly passed** from `tl.py` into `run_permutation_test` (no more hidden reliance on adata.var after the fact).
+
+### Post-review hardening (2026-06-27)
+- **Fragility fix (high)**: Added explicit schema validation + safe column access + clear warning in `_run_memento_de` for expected 'de_coef'/'de_pval' from memento.binary_test_1d. Prevents silent fallback (or crash) if upstream memento-de changes column names/structure. Updated optional dep pin to "memento-de>=0.1.0,<0.3.0".
+- **Cleanup + consistency (high)**: `run_permutation_test` now owns the small-space FDR decision (use_fdr=False + disabled_reason="small_permutation_space" when n_perm < 100). Removed vestigial dead `if not perm_use_fdr` and useless max_perm check/assign in tl.py. `disabled_reason` (and `perm_disabled_reason`) now properly returned and stored in adata.uns["scatrans"] metadata.
+- **Diagnostics improvement**: MixedLM per-gene fits now count genes hitting the neutral-except fallback (`n_genes_failed_fit`). Recorded in diagnostics["mixed_model"] (active_score) and DE metadata (differential_expression). Warning emitted when >0 so users are not unaware of silent neutral (delta_var=0, p=1) genes.
+- Only confirmed high-impact issues from the review list were addressed; medium/nuance items (e.g. EB small-ref prior, advanced fallback diag completeness) were already mitigated by existing diagnostics/fallbacks or not correctness bugs.
+
+All new issues addressed. Full test suite green.
+
+## [0.9.2] - 2026-06-20
+
+### Added / Changed
+- `filter_active_genes` now accepts `logfc_direction="up"|"down"|"both"` (default remains `"up"` for backward compatibility and "active" semantics).
+  - `logfc_cutoff` is interpreted as a positive magnitude in all modes.
+  - `"down"`: selects logFC < -cutoff (downregulated genes from differential_expression results).
+  - `"both"`: selects |logFC| > cutoff.
+  - Sorting for pure-DE tables is now direction-aware (most-negative-first for down, largest |logFC| for both).
+  - This directly supports the common request for downregulated candidates in standalone DE workflows.
+- Updated docstrings, tests, and examples.
+
 ## [0.9.0] - 2026-06-19
 
 ### Added

scatrans-0.9.4.2.dev0/CITATION.cff

@@ -0,0 +1,20 @@
+cff-version: 1.2.0
+title: "scATrans: Single-cell Active Transcription Analysis"
+message: >-
+  If you use this software, please cite it using the metadata below.
+type: software
+authors:
+  - name: "scATrans Developers"
+repository-code: "https://github.com/scATrans/scatrans"
+url: "https://github.com/scATrans/scatrans"
+license: Apache-2.0
+version: 0.9.2
+keywords:
+  - single-cell
+  - RNA-seq
+  - unspliced RNA
+  - nascent RNA
+  - active transcription
+  - bioinformatics
+  - differential expression
+  - gene set enrichment

{scatrans-0.9.2.dev0 → scatrans-0.9.4.2.dev0}/MANIFEST.in

@@ -5,7 +5,13 @@
 include LICENSE
 include README.md
 include CHANGELOG.md
+include CITATION.cff
 include pyproject.toml
+include conftest.py
+include conftest_fixtures.py
+
+# Include the full test suite so that `pytest` can be run from sdist/unpacked source
+graft tests
 
 # Include the GitHub Actions workflows (requested)
 include .github/workflows/ci.yml