Claude Skills by HolobiomicsLab

Use when when working with mass spectrometry imaging data from metabolites treated with derivatizing matrices (e.

Use when use when the workflow requires descriptor-fingerprint-feature-engineering.

Use when when you have a trained BitterPredict classifier, a dataset of molecules with computed descriptors and known bitter/not-bitter labels, and want to understand which descriptor categories (e.g., molecular weight, lipophilicity, topological, pharmacophoric) drive prediction decisions.

Use when when you have calculated molecular descriptors from validated chemical structures and need to prepare them for BitterPredict.m classification. Use this skill after descriptor calculation is complete and before invoking BitterPredict.m for bitter/non-bitter predictions.

Use when when you have cloned a Python package repository and need to prepare a working environment for development, debugging, or contribution. Specifically when the package declares dev dependencies in setup.py or pyproject.toml and maintains a pytest test suite in a tests/ directory.

Use when you have computed high-dimensional embeddings (e.

Use when after generating large feasible flux distributions (e.g., 1 million sampled solutions per cell line) from constrained metabolic models, apply t-SNE when you need to assess whether distinct biological samples (e.

Use when when you have intermediate JSON data that must be selectively transformed or enriched according to declarative conversion rules—for example, when extracting experimental metadata from tabular spreadsheets, you need to map certain fields to computed or filtered values, apply conditional.

Use when you have a sparse pairwise distance matrix derived from nearest neighbor indexing of MS/MS spectra (or similar high-dimensional objects) and need to partition spectra into groups based on local density and neighborhood connectivity.

Use when you have normalized peak intensity data (or absence/presence matrices) from metabolomics experiments with multiple samples and need to quantify compositional differences between them prior to multivariate analysis. This step is essential when testing whether categorical factors (e.

Use when when preparing a software release, testing contribution workflows, or auditing package availability: verify that matchms can be installed and imported successfully from all advertised distribution channels (PyPI and Bioconda) to confirm the package metadata, dependencies, and entry points.

Use when you have loaded a collection of molecular fingerprint vectors (such as biosynfoni fingerprints from a deposited dataset) and need to characterize their statistical and distributional properties before using them for machine learning, similarity searching, or method validation.

Use when you have normalized peak-abundance matrices with sample metadata containing categorical treatment variables (e.

Use when when a research software project is decomposed into distinct subproject components (e.

Use when when you have deployed NP Classifier using Docker Compose and need to confirm that both the server and TensorFlow Serving containers are running and healthy before sending SMILES strings to the /classify endpoint.

Use when when you have .mzML or .abf LC-HRMS raw data files that require MS-DIAL-based feature detection, chromatogram alignment, and metabolite identification, and you need to ensure reproducibility across local machines, cloud, and HPC systems without manual tool installation and dependency.

Use when you have a vendor mass spectrometry raw file (e.g., .raw format) that requires conversion to Aird format using AirdPro CLI, and you are running on macOS or Linux.

Use when when you need to confirm that a published Docker image (e.g., hosted on Docker Hub) can be pulled and instantiated successfully, and when the target tool has a defined entry point or CLI interface.

Use when after building multiple Docker image variants (e.g., cli, dev, linux, windows) using multi-stage builds with --target flags, and you need to verify that each variant's size falls within documented ranges (e.g., cli 6–7 GB, dev 9–11 GB, linux 8–10 GB, windows 4–5 GB).

Use when you need to containerize a C#-based Windows application (like AirdPro CLI) for Linux deployment, require Wine and .NET Framework 4.

Use when a mature scientific package (e.g., Mummichog 3) is being migrated to a new GitHub organization that enforces standardized project structure, and the current setup.py, pyproject.toml, requirements.txt, .

Use when you have access to a multi-component research software repository (e.

Use when when a user submits one or more MS/MS spectra and has declared or implied a domain context (microbial, plant, tissue, microbiome, food, or metadata aggregation), and the search must be executed against the appropriate domain-curated spectral library.

Use when when you need to enable non-programmers or domain experts to formulate complex, unambiguous queries over specialized data (e.

Use when when you have just loaded the rawrr R package and need to confirm that the bundled .NET 8.0 assembly (rawrr.exe) is present and functional before performing any mass spectrometry data extraction operations.

Use when you need to containerize a C# application (e.g., AirdPro) that targets .NET Framework 4.8 and must run on Linux hosts via Docker, but the application was originally built for Windows. Use this skill when you are building multi-stage Docker images from a Ubuntu 22.

Use when when training a deep neural network on mass spectrometry spectral data where overfitting is a risk (especially with data augmentation applied), and when you need both regularization during training AND uncertainty quantification at inference time via multiple forward passes with dropout.

Use when you have prototyped a novel data-dependent acquisition strategy that uses dynamic exclusion with intensity or ROI weighting, and you need to quantitatively compare its MS/MS coverage and intensity performance against a simpler baseline (TopN) before testing on real instrumentation.

Use when when building an automated converter discovery and job enumeration system where converter classes are dynamically loaded from package directories and you need to extract and validate their internal conversion method signatures without prior knowledge of which converters will be available.

Use when after abundance-correlation-based feature group refinement when you observe that larger feature groups (particularly those with 3+ features in the same m/z–retention-time window) may contain features with different peak shapes or retention-time shifts in their EICs, or when abundance.

Use when after training or loading a pre-trained deep learning model that produces high-dimensional embeddings (e.g., 200-dimensional vectors from MS2DeepScore). Use this skill when you have a set of spectra with known chemical class annotations (e.

Use when after instantiating and invoking a sinusoidal formula embedding layer (such as SCARF embeddings in MIST-CF) on chemical formula inputs, validate that the output embeddings meet dimensionality and value constraints before using them for downstream transformer or ranking tasks.

Use when after feature detection and quality control have produced a feature table in TSV format from Asari or equivalent preprocessing.

Use when after feature table normalization and imputation are complete, immediately before MS1 and MS2 annotation.

Use when you have 512-dimensional representation vectors output from paired ResNet18 encoders processing augmented ion images, and you need to: (1) introduce an intermediate projection space to enable contrastive loss optimization without trivial/collapsed solutions, (2) further compress learned.

Use when after generating combined or alternative scores for a set of BGC-metabolite (GCF-MF) link candidates, you need to evaluate whether a scoring function preferentially ranks true validated links higher than spurious ones.

Use when after installing ENPKG or any component of the workflow via conda/pip dependency manifests and before executing workflow scripts.

Use when when extending a multi-service project (like MAGMa with its four subproject components) to container orchestration, and you need to ensure each microservice (magmaweb, joblauncher, job, pubchem) receives the correct configuration—such as port mappings, service URLs, and data paths—without.

Use when when implementing or auditing a data replacement method (e.g., `mz<-`, `intensity<-`) in an MsBackend subclass that must enforce ordering or format constraints on peak data.

Use when you have predicted retention times from one or more machine learning models (DNN, Gaussian Process, or ensemble) applied to small-molecule chromatography data, along with corresponding experimental ground-truth retention times, and need to quantify prediction accuracy and rank competing.

Use when when a user uploads a JSON project document to the Pairing Omics Data Platform and you need to determine whether it satisfies the platform's data structure requirements, including all mandatory fields, proper data types, and constraint satisfaction for paired omics metadata (e.

Use when building or maintaining a system that fetches metadata from multiple independent external web services and needs to diagnose why annotation runs fail or slow down. Use it specifically when you need to distinguish between service-level failures (e.

Use when you are developing or comparing new data-dependent acquisition (DDA) strategies in ViMMS and need to evaluate how well each strategy fragments sampled compounds from the HMDB database.

Use when you have tandem MS/MS spectra annotated by at least two of GNPS (FBMN), ISDB-LOTUS (CFM-ID 4.0 spectral matching), and Sirius 6, and you need to rank features by annotation agreement rather than trust a single tool's output.

Use when after loading raw Agilent Unknowns Analysis CSV output and when you need to convert tentative compound identifications (matched only by GC-MS library cosine similarity or Match.Factor score) into searchable, curated chemical records with exact masses and multi-source confirmation.

Use when you have measured execution times from multiple scripts that exercise different combinations of categorical variables (e.g., plot types: chromatogram, mobilogram, peakmap, peakmap-marginals, spectrum, subplots;

Use when before initiating raw file conversion or feature extraction, when you have a heterogeneous collection of raw LC-MS files (.raw or .mzML) and sample information scattered across instrument logs, sequence files, or spreadsheets.

Use when you have a published scientific article describing a computational method (e.g., natural products annotation, MS/MS data processing) and need to assess reproducibility before expert scrutiny.

Use when you have a tabular file (CSV or Excel) that has been manually or semi-automatically tagged with export tags, and you need to verify tag correctness before running the extract command to convert the tagged table into intermediate JSON.

Use when your project JSON document contains genome identifiers but lacks organism name or taxonomic annotations. The platform needs to auto-populate these fields to enable browsing and cross-linking with public genomic databases. Trigger this skill when you have genome IDs (e.