Repository navigation
test(differential): the absence-path ml fit needs fittable counts - #176
Merged
Merged
Conversation
The glmGamPoi-absence testset asserted all(r.status .== "ok") on a 4-sample matrix of small counts. MEASURED with MASS::glm.nb (R 4.5, MASS 7.3-60): that data cannot reach status ok -- one taxon's theta runs to its upper bound (status boundary) and the other's alternating fit produced NaNs (status failed). The test's CI runs were cancelled before completing on the feature branch, so the first completed run on main (da9754f) was the first to expose it. The test now fits the injected-effect dataset with a TSS offset -- the dataset the "Negative-binomial fit recovers an injected effect" testset already proves fits -- keeping the assertion's intent: the requireNamespace override defers to base, so the default dispersion still fits. Generated by Mistral Vibe. Co-Authored-By: Mistral Vibe <vibe@mistral.ai>
|
Note Currently processing new changes in this PR. This may take a few minutes, please wait... ⚙️ Run configuration
📒 Files selected for processing: 1
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What this change does
The "nb_glm without glmGamPoi refuses with GlmGamPoiUnavailable" testset now fits the injected-effect dataset (with a TSS offset) for its final assertion — the ml-path fit with the requireNamespace override in place — instead of the minimal 4-sample matrix.
Why
The test asserted
all(r.status .== "ok")on[5 6; 7 3; 9 4; 4 8]. MEASURED locally with MASS::glm.nb (R 4.5.3, MASS 7.3-60): that data cannot reach status ok — one taxon's theta runs to its upper bound (status boundary, "the fit is effectively Poisson") and the other's alternating fit produced NaNs (status failed, "the fit did not converge"). The feature branch's CI runs were cancelled before completing, so the first completed run on main (da9754f) was the first to expose it. The dataset "Negative-binomial fit recovers an injected effect" already proves fits; with it, all three taxa fit ok with the override in place (MEASURED locally: statuses ["ok", "ok", "ok"]).The assertion's intent is unchanged: the override defers to base for every other package, so the default dispersion still fits normally with it in place.
How to run its tests
R_HOME=… julia --project=. test/unit/test_differential.jl— the affected testset is the last one. Needs R and MASS; skips loudly without them.What it leaves out
No production code changes; the fit machinery is untouched. Local runs without zCompositions show unrelated failures in the "Route" testset (it needs zCompositions); CI has it and that testset passed on da9754f.
Versions tested against
Julia 1.13.1, R 4.5.3 (conda-forge) with MASS 7.3-60.
Generated by Mistral Vibe.
Co-Authored-By: Mistral Vibe vibe@mistral.ai