Fix use count based fp

[thomasATbayer]

https://github.com/Bayer-Group/MotherML/blob/main/src/mother/feature_generation/core.py https://github.com/Bayer-Group/MotherML/blob/main/src/mother/feature_generation/config.py currently the Morgan Fingeprint setting in Mother is to not use counts. Binary fingerprints are usually not a good choice for ML, so should make sure that the standard setting is to use counts. This also enables better treatment in the case of bit colission, , as they at least increase the count by +1 instead od just ignoring. And maybe increase fpsize to 2048 in core.py too asis already in the config. Most users will not change on their own, and for some tasks using binary FP can make models perform substantially worse, which is also backed by literature: Count-Based Morgan Fingerprint: A More Efficient and Interpretable Molecular Representation in Deve… Count your bits: fingerprint benchmarking to assess broad chemical space representation | Journal o…

[Copilot]

Pull request overview

This PR updates the default fingerprint-generation behavior to use count-based fingerprints by default, and adjusts Morgan fingerprint defaults to better align with the framework’s configured fingerprint parameter defaults.

Changes:

• Switch default use_counts to True for FingerprintsGeneric and MorganFingerprints.
• Increase MorganFingerprints default fpSize to 2048.
• Switch FeatureGenerationConfig.use_counts default to True, affecting the default feature-generation pipeline behavior.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
src/mother/feature_generation/core.py	Changes default fingerprint behavior (count vs. binary) and Morgan fingerprint default size/params.
src/mother/feature_generation/config.py	Changes the default config setting that controls whether count fingerprints are used in the feature-generation pipeline.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.

Suppressed comments (1)

src/mother/feature_generation/core.py:70

• The public docs still state that fingerprints default to binary output (use_counts=False) and show MorganFingerprints examples with fpSize=1024. With this PR changing the defaults to use_counts=True (and MorganFingerprints fpSize=2048), mkdocs/docs/feature_generation.md becomes misleading and should be updated to reflect the new default behavior (and ideally update the examples/table accordingly).

    def __init__(self, fp_type: str, parameters: dict, use_counts: bool = True) -> None:
        self.fp_type: str = fp_type
        self.parameters: dict = parameters
        self.use_counts: bool = use_counts

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

Suppressed comments (2)

src/mother/ml/models/m_tabpfn.py:386

• Same scope concern as above: this PR is about defaulting Morgan fingerprints to count-based, but it also changes TabPFNClassifierMother defaults by setting inference_precision=torch.float32. Please either document this in the PR description (and any user-facing docs) or move it to a separate PR.

        #   TabPFNClassifier.create_default_for_version(ModelVersion.V2)
        if "model_path" not in kwargs:
            kwargs["model_path"] = TabPFNClassifier.create_default_for_version(ModelVersion.V2).model_path
        kwargs.setdefault("inference_precision", torch.float32)

src/mother/feature_generation/config.py:143

• The default use_counts has been changed to True here, but the user documentation still states that fingerprints are binary by default and shows examples using MorganFingerprints(..., fpSize=1024) (see mkdocs/docs/feature_generation.md:34-39, 50, 91-92). This will mislead users after this change; please update the docs/examples to reflect count-based defaults and the 2048-bit Morgan default (and optionally mention the breaking-behavior change).

    fingerprints: List[Dict[str, Any]] = Field(default=[], description="List of fingerprint generator settings")
    maccs: bool = Field(default=False, description="Flag if maccs fingerprints should be generated")
    chemical_descriptors: Optional[ChemicalDescriptorsParams] = Field(default=None)
    use_counts: bool = Field(default=True, description="Whether to use count fingerprints")

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.

Suppressed comments (2)

Previously missed (1) — in code that hasn't changed since the last review.

test/unit/test_use_counts.py:129

• The PR also changes MorganFingerprints default fpSize to 2048, but the “default” test currently passes fpSize=2048 explicitly. That means a regression in the default size wouldn’t be caught by this suite.

    def test_morgan_count_default(self, mols_with_repeated_substructures):
        """MorganFingerprints with default use_counts=True should produce count output."""
        fg = MorganFingerprints(radius=2, fpSize=2048)
        fg.fit()

src/mother/feature_generation/core.py:132

• MorganFingerprints.__init__ accepts **kwargs (and callers pass many generator options via MorganFingerprintsParams().model_dump()), but those kwargs are currently ignored. As a result, settings like countSimulation, useBondTypes, includeChirality, etc. have no effect when using the MorganFingerprints convenience class, which is especially problematic now that count-based fingerprints are the default.

    def __init__(
        self, radius: int = 2, fpSize: int = 2048, include_chirality: bool = False, use_counts: bool = True, **kwargs
    ) -> None:
        super().__init__(
            "MorganFP",
            {
                "radius": radius,
                "fpSize": fpSize,
                "includeChirality": include_chirality,
            },
            use_counts=use_counts,
        )

[SommerKai]

closes #75

[Copilot]

🟡 Changes recommended

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Review details

• Files reviewed: 5/6 changed files
• Comments generated: 1
• Review effort level: Lite

[Copilot]

🟡 Changes recommended

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Review details

• Files reviewed: 5/6 changed files
• Comments generated: 1
• Review effort level: Lite