Conversation
Bam comp fix
Refinement
Planemo tests
Rust tests
build: cargo deps overhaul build: cap htslib/rust-htslib -> build issues test: update pytests to accomodate new mpl test: dependabot for versions test: pin mpl version to avoid hitting png differences build: minimal python 3.12
test: fix planemo tests under latest MPL test: planemo test to galaxy version 26 ci: boost action runner versions
Salvaged from #1369: keeps only the computeMatrix test, which passes against the current Rust bindings. The alignmentSieve, bamCompare and bamCoverage tests from that PR were dropped because their Rust function signatures have since changed (and alignmentSieve2 is not yet wired up on 4.0.0). Co-authored-by: gerikson <galina.erikson@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Both existing tests only compared the rendered PNG. Adds numeric assertions on the tabular outputs, filling gaps found while triaging the stalled #1372/#1374 tests: - plotPCA: verify --outFileNameData (the existing test passed this flag but the check was commented out). Asserts the sign-independent eigenvalue column and table shape. - plotFingerprint: verify --outQualityMetrics + --JSDsample (AUC and JS Distance), neither previously tested anywhere. Expected values derived by running the current tools; verified with `pixi run pytest`. Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
…t/strand filtering (#1434) Fills coverage gaps identified while triaging the stalled #1369 tests: - bamCoverage: --normalizeUsing RPKM (RPKM was untested; suite only had CPM/RPGC) and --scaleFactor. - alignmentSieve: --minFragmentLength/--maxFragmentLength and --filterRNAstrand (neither was tested anywhere). Expected values derived by running the current tools; all pass under `pixi run pytest`. Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* Compute PCA with scipy instead of scikit-learn Replace the sklearn PCA and StandardScaler in Correlation.plot_pca with a scipy/numpy SVD implementation, reproducing sklearn's output (column standardization with population std, deterministic svd_flip sign convention, explained_variance_ from singular values). Drop the now-unused pandas and scikit-learn dependencies. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Fix transpose, log2/rowCenter, and small-ntop bugs in plot_pca; add PCA tests The scipy/SVD PCA rewrite inherited three pre-existing plot_pca bugs from the sklearn version: - --transpose crashed with a shape mismatch (np.dot(m, Wt.T)); U*S already gives sample projections, so orient as (components, samples) via Wt.T. - --log2 / --rowCenter were no-ops: they mutated self.matrix after m had been copied during ntop filtering. Now applied to a float copy before variance filtering. - --ntop below the sample count crashed the scatter with IndexError; guard with a clear sys.exit instead. Adds test_plotPCA.py coverage (sign-invariant coordinate/eigenvalue regressions, ntop behavior, transpose, CLI validation exits) that passes against both the sklearn and scipy implementations. * Make plotPCA coordinate test portable across BLAS backends test_plotPCA_default_coordinates failed on macOS CI: after PC1 the untransposed eigenvalues are near-degenerate, so the eigenvectors rotate freely and np.argpartition breaks top-ntop variance ties differently on Accelerate vs OpenBLAS, making per-feature coordinates non-reproducible. Replace it with test_plotPCA_default_eigenvalues, asserting only the portable eigenvalues; coordinate-level regression stays covered by the well-separated transpose case (test_plotPCA_transpose). --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The second plotPCA test compared the PCA tabular output with an exact diff. Its duplicate-sample column is mathematically zero and comes out as tiny, BLAS/platform-dependent noise (~1e-18), which failed the diff in CI. Replace the exact compare with assert_contents pinning structure, the stable PC1 coordinates, and the leading eigenvalue.
* boost rust edition, pin pyo3 together with minimal python version * fix set_bad deprecation mpl * set mpl defaults * alignmentsieve rust implementation * drop hardcoded /tmp in pytests in favor of tempfile * cramfile ref included for pytest/planemo test --------- Co-authored-by: Mohit Navandar <133376879+NavandarM@users.noreply.github.com>
some optimizations on computematrix / mbs start exhaustive testing of computematrix module - reference point collect computematrix tests in one place
- Add zizmor pre-commit hook (.pre-commit-config.yaml) - Pin all GitHub Actions to commit SHAs (unpinned-uses) - Add persist-credentials: false to checkout steps (artipacked) - Add minimal top-level permissions to workflows (excessive-permissions)
* tests * Fix stale plot baselines, support matplotlib 3.10.5+ Several baseline PNGs weren't regenerated after the dpi/rcParams change in matplotlib_defaults.py, breaking image comparison tests even on the pinned mpl 3.11. Regenerated them and bumped tolerances for cross-version antialiasing drift. Loosened the mpl pin from ==3.11 to >=3.10.5,<3.12 since 3.10.0-3.10.4 have a Path.__deepcopy__ RecursionError on Python 3.14. * Regenerate plotHeatmap ggplot baseline for matplotlib 3.11.1 bbox_inches='tight' makes the --ggplot heatmap's saved size depend on exact text/legend extents, which shifted by 15px in height between 3.11.0 and 3.11.1. CI resolves the newest matching matplotlib via pip (3.11.1), so pin the baseline there. Pixel dimension mismatches can't be absorbed by tolerance since compare_images requires exact size. * Regenerate stale Galaxy wrapper test baselines Same root cause as the pytest suite: the matplotlib_defaults.py dpi rewrite changed every plot's pixel output, but galaxy/wrapper/test-data baselines were never regenerated. planemo's sim_size comparison was failing size deltas of roughly 2x on plotPCA, plotCorrelation, plotCoverage, plotEnrichment, plotFingerprint, plotHeatmap, and bamPEFragmentSize. Regenerated all of them by replaying each tool's <test> block CLI args directly. * Widen Galaxy wrapper sim_size deltas for remaining plot tests plotCorrelation, plotCoverage, plotEnrichment, and plotHeatmap still failed planemo's byte-size comparison after the baseline regen: CI's actual output is consistently 5-15% larger than what's reproducible locally, in a way that doesn't respond to CLI-arg fixes (some outputs are byte-identical before and after). Two separate CI runs produced identical actual byte counts, so this is a stable environmental difference (matplotlib patch version), not flakiness - widen delta to give it headroom rather than chase an exact match. Also corrected plotCorrelation's test invocation to use the real Galaxy default corMethod (spearman, not pearson). * Drop bbox_inches='tight' from plotHeatmap savefig The figure already has an explicit figsize and constrained_layout, so the tight-bbox recrop was redundant - and it was also the actual source of the plotHeatmap_ggplot size drift between matplotlib patch versions (finding C), since it recomputes the crop from rendered text/legend extents that shift slightly release to release. Confirmed output dimensions are now identical (393x2519) under both 3.11.0 and 3.11.1, so pixi's conda-resolved matplotlib and CI's pip-resolved matplotlib no longer need to match exactly for this test to pass. Regenerated the four affected baselines (pytest + galaxy wrapper). * Widen plotHeatmap Galaxy wrapper deltas after bbox_inches fix Dropping bbox_inches='tight' changed the output size for both heatmapper tests (not just the ggplot one), so result2's previously- fine delta now also needs headroom for the same environmental rendering drift documented for the other Galaxy wrapper tests. * Drop argparse.FileType in favor of path validation + explicit open() Python 3.14's argparse.FileType.__init__ now emits a PendingDeprecationWarning on every construction, which fired once per parser build across the test suite (58 warnings). Every FileType usage in this codebase only ever consumed the '.name' attribute or passed the handle straight into a single write call, so none of it needed an eagerly-opened handle at parse time. Added parserCommon.readableFile alongside the existing writableFile as a path-validating argparse type, switched all 8 FileType(...) call sites to the appropriate one, and open() explicitly at the point of use for the handful of writes that need a real file object (save_BED). Down to 2 warnings suite-wide (an unrelated scipy one).
exhaustive testing for multibamsummary, and computematrix (scale-regions/refpoint). Edge cases (chromosome boundaries, filtering, blacklist) taken care of.
* gz bed/gtf and test * gz blacklist testcase * test matrix planemo * planemo test defaults * include channels in planemo action * include panic message over bare unwraps
* update changelog * mend * docs orphan for warnings * drop doc reqs, fix python docstring * drop unused code from new entrypoints * point rtd to docs req * drop mpl from action reqs * action to test docs * minimal versions, include lock * centralize version check deeptools/galaxy, rust version in docs/deeptools * include authors * ensure rust in rtd yaml is an allowed rust version * cleanup readme, https * docs overhauls, remove plotly refs, update parser * rust 1.87 minimum * ignore venv * move pytest action over to pixi * pip as a pixi dep * of course dist upload needs mussllinux builds * set stack min for osx build in pixi env * up stack * 100MB stack for osx build ? * strip wheels, cap rust version to avoid conda-forge rust build error for osx * https in docs, clarify installation instructions, delete rst
This was referenced Sep 5, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Related issues:
Closes #1423, gzip support included for sort arm
Closes #1422, should be fixed (though not explicitly tested)
Closes #1421, single output stream for alignmentSieve
Closes #1384, fixed
Closes #1333, plotly backend is deprecated
Closes #1195, Closes #662, performance optimized
Closes #1194, regions should be respected strictly now
Closes #1180, Closes #1093, Closes #1088, alignmentSieve works with new back-end
Closes #1144, closes #1030
Closes #1140, multiprocessing no longer used in computeMatrix
Closes #1108, Closes #1130 fixed
Closes #1094, no longer relevant
Closes #1074, no longer relying on pysam
Closes #955, no longer relevant.
Closes #729