Source: https://github.com/aipoch/medical-research-skills

Matchms Skill

When to Use

• Use this skill when you need process, clean, and compare mass spectrometry (ms/ms) spectra with matchms; use when you need reproducible spectral filtering and similarity scoring for metabolomics workflows in a reproducible workflow.
• Use this skill when a data analytics task needs a packaged method instead of ad-hoc freeform output.
• Use this skill when the user expects a concrete deliverable, validation step, or file-based result.
• Use this skill when scripts/similarity_pipeline.py is the most direct path to complete the request.
• Use this skill when you need the matchms package behavior rather than a generic answer.

Key Features

• Scope-focused workflow aligned to: Process, clean, and compare mass spectrometry (MS/MS) spectra with Matchms; use when you need reproducible spectral filtering and similarity scoring for metabolomics workflows.
• Packaged executable path(s): scripts/similarity_pipeline.py.
• Reference material available in references/ for task-specific guidance.
• Structured execution path designed to keep outputs consistent and reviewable.

Dependencies

• Python: 3.10+. Repository baseline for current packaged skills.
• Third-party packages: not explicitly version-pinned in this skill package. Add pinned versions if this skill needs stricter environment control.

Example Usage

cd "20260316/scientific-skills/Data Analytics/matchms"
python -m py_compile scripts/similarity_pipeline.py
python scripts/similarity_pipeline.py --help

Example run plan:

1. Confirm the user input, output path, and any required config values.
2. Edit the in-file CONFIG block or documented parameters if the script uses fixed settings.
3. Run python scripts/similarity_pipeline.py with the validated inputs.
4. Review the generated output and return the final artifact with any assumptions called out.

Implementation Details

• Execution model: validate the request, choose the packaged workflow, and produce a bounded deliverable.
• Input controls: confirm the source files, scope limits, output format, and acceptance criteria before running any script.
• Primary implementation surface: scripts/similarity_pipeline.py.
• Reference guidance: references/ contains supporting rules, prompts, or checklists.
• Parameters to clarify first: input path, output path, scope filters, thresholds, and any domain-specific constraints.
• Output discipline: keep results reproducible, identify assumptions explicitly, and avoid undocumented side effects.

1. When to Use

Use this skill when you need to:

• Import and harmonize MS/MS spectra from common community formats (e.g., MGF/MSP) before analysis.
• Clean spectra (peak filtering, intensity normalization) to improve downstream similarity scoring and identification.
• Compute spectral similarity (Cosine/Modified Cosine/Fingerprint-based) for library matching or clustering.
• Build reproducible, configurable processing pipelines for metabolomics projects.
• Compare many spectra efficiently (all-vs-all or query-vs-library) and store/inspect score outputs.

2. Key Features

• Import/Export support: Read spectra from mzML, mzXML, MGF, MSP, and JSON (depending on installed readers).
• Filtering & harmonization: Metadata standardization, peak cleaning, intensity normalization, and other reusable filters.
• Similarity scoring:
  • Cosine similarity (Greedy/Hungarian variants)
  • Modified Cosine (accounts for precursor mass shifts)
  • Fingerprint-based similarities (when molecular fingerprints are available)
• Pipeline composition: Chain filters and scoring steps into repeatable workflows.

Additional reference material (if present in the repository):

• Filters: references/filtering.md
• Similarity: references/similarity.md
• Workflows: references/workflows.md

3. Dependencies

• matchms (version depends on your environment; pin in your project, e.g., matchms>=0.20,<1.0)
• numpy (e.g., numpy>=1.20)
• scipy (e.g., scipy>=1.7)
• rdkit (optional; required for chemistry/fingerprint-related functionality, version varies by distribution)

4. Example Usage

A minimal, runnable example that loads spectra from an MGF file and computes pairwise cosine scores:

from matchms.importing import load_from_mgf
from matchms import calculate_scores
from matchms.similarity import CosineGreedy

def main():
    # Load spectra from an MGF file
    spectra = list(load_from_mgf("data.mgf"))

    # Compute similarity scores (all-vs-all)
    scores = calculate_scores(
        references=spectra,
        queries=spectra,
        similarity_function=CosineGreedy()
    )

    # Iterate over computed scores
    for (reference_idx, query_idx, score, n_matches) in scores:
        print(
            f"ref={reference_idx:>3} query={query_idx:>3} "
            f"cosine={score:.4f} matches={n_matches}"
        )

if __name__ == "__main__":
    main()

5. Implementation Details

• Data model: Matchms operates on Spectrum objects containing peak m/z and intensity arrays plus metadata (e.g., precursor m/z, charge, compound name/identifier).
• Filtering stage: Typical pipelines apply filters to:
  • standardize/repair metadata fields,
  • remove noise peaks (e.g., by intensity threshold or m/z window rules),
  • normalize intensities (commonly to a maximum of 1.0 or to unit norm). See references/filtering.md for filter patterns and recommended sequences.
• Cosine similarity (Greedy/Hungarian):
  • Peaks are matched within an m/z tolerance (implementation-specific defaults; configure via the similarity class parameters).
  • Greedy matching selects best available peak matches iteratively.
  • Hungarian matching solves an assignment problem to maximize total match score under one-to-one constraints.
• Modified Cosine:
  • Extends cosine matching by allowing peak alignment with a precursor mass shift, improving matching for related compounds/adducts.
  • Typically requires precursor m/z metadata to be present and consistent.
• Fingerprint similarity (optional):
  • Requires molecular fingerprints (often derived via RDKit) and compares spectra/compounds using fingerprint similarity metrics.
  • Use when you have structure annotations or can compute fingerprints reliably.
• Workflow reproducibility:
  • Prefer explicit, ordered filter chains and pinned dependency versions.
  • Store configuration (tolerances, normalization choices, filters used) alongside results for traceability. See references/workflows.md for pipeline organization guidance.