Claude Skills by HolobiomicsLab

Use when you have a quantification table (rows = metabolite features, columns = samples with abundance values), corresponding metadata table (sample annotations, groupings), and spectral data files, and you need to produce a unified JSON dashboard artifact that can be loaded into an interactive.

Use when you have a MemoMatrix (sample-by-fingerprint matrix) from aligned MS2 spectra and need to visually compare sample similarity or clustering patterns, especially when samples show poor feature overlap, strong retention time shifts across different LC methods, or were acquired on different.

Use when after running a ViMMS Environment simulation with save_eval flag enabled, when you need to preserve the link between each simulated MS/MS scan in the output mzML file and its source chemical definition, fragmentation parameters, and evaluation metrics for later analysis, comparison, or.

Use when you have completed peak picking with two or more competing tools (e.g., IDSL.IPA, MZmine 2, xcms, MS-DIAL) on the same LC/HRMS dataset(s) and need to quantify which performs better. Use this skill when tool selection claims require validation (e.g., 'IDSL.

Use when you have a Galaxy installation (specifically Galaxy Master branch commit c429777c93680dcee449fe410f5360afbe673758 or compatible) and need to add metabolomics analysis capabilities including tools for XCMS integration, mass spectrometry file reading (via MSFileReader), and metabolite.

Use when when you have LC-MS/MS acquisitions in DDA mode and need to train a customized DNMS2Purifier model to purify chimeric MS/MS spectra specific to your experimental conditions, metabolite classes, or ionization settings.

Use when when you have read LC-MS peak table data from Excel (or equivalent) into R and need to organize it into a structured object that tracks feature abundances, sample information (injection order, QC status), and feature metadata (mass, retention time, Feature_ID) simultaneously.

Use when after data merging and before applying batch correction algorithms (ComBat, SVA, or normalization techniques) to a merged feature table from non-targeted LC-MS/MS metabolomics data.

Use when you have raw mass spectrometry files in one of the supported formats (mzML, mzXML, msp, metabolomics-USI, MGF, JSON) and need to parse out metadata fields (e.

Use when you have a feature table and accompanying CSV metadata that includes a 'Sample Type' field (or equivalent) with entries such as 'BLANK', 'QC', 'STD', or 'Unknown'.

Use when reading mass spectral library files (particularly MoNA EI or MS2 libraries) where structural metadata like SMILES information is embedded in general-purpose fields (e.g., Comment field) rather than in the dedicated SMILES field expected by mspcompiler's downstream processing steps.

Use when immediately after importing raw mass spectrometry data from mzML, mzXML, msp, MGF, or JSON formats into matchms.

Use when you have completed independent batch searches across one or more domain-specific MASST tools (microbeMASST, plantMASST, tissueMASST, microbiomeMASST, foodMASST) and received multiple separate output files (_microbe.html, _plant.json, _matches.tsv, _library.tsv, _datasets.tsv, _count_domain.

Use when when you have preprocessed MS/MS spectra from multiple source repositories or instruments with inconsistent metadata field naming, formats, or values (e.g., mixed adduct notations like '[M+H]+' vs '[M+H]⁺', variable instrument type strings, or non-standard collision energy units).

Use when after processing LCMS feature data through Blueshift or Gravity but before finalizing results.

Use when you have multiple CSV feature lists from different acquisition methods (e.g., LC-MS vs LC-IMS-MS) or processing software, each using different naming conventions, retention time scales, or m/z precision;

Use when when a tool like TARDIS extends its API to accept multiple input types (e.g., both file paths and MsExperiment objects), and you need to confirm that screening-mode diagnostic outputs (e.g., EIC plots, peak detection metrics) are identical regardless of which invocation pattern is used.

Use when you have raw or MZmine-processed MGF files (containing MS/MS spectra with m/z values, intensities, and precursor masses) that need to be segmented by sample, deduplicated, or prepared for fragment annotation and adduct analysis in the MolNotator pipeline.

Use when when processing MGF-format MS2 spectral libraries (e.g., GNPS) that contain SMILES but lack the Molecular Formula field, and you need to prepare the library for MS-DIAL import or polarity-based separation workflows.

Use when you have paired microbiome (16S or metagenomic taxonomy/functions at genus or finer level) and metabolome (LC-MS or similar profiled metabolites) data from the same samples and want to: (1) predict unobserved metabolite abundances from microbiome composition;

Use when when you have raw count matrices from paired microbiome (16S rRNA or metagenomic) and metabolomic (LC-MS/MS) profiling data that will be used to train or apply a predictive model (e.g., MiMeNet, MelonnPan, Random Forest) to predict metabolite abundances from microbial composition.

Use when you have paired microbiome and metabolomic abundance tables (samples × features) with relative abundance or raw count values, and you are preparing data for downstream regression or neural network modeling of microbe-metabolite relationships.

Use when when starting with raw paired microbiome (16S rRNA, metagenomic taxonomic or functional features) and metabolome (LC-MS/MS, NMR) count tables from the same biospecimens, and planning to train prediction models or co-abundance networks.

Use when you have paired microbiome (16S rRNA, metagenomic) and metabolomic (LC-MS, GC-MS) abundance tables from the same biosamples, and you want to predict which metabolites are recoverable from microbial composition alone and identify groups of microbes and metabolites with correlated.

Use when you have OnDiskMSnExp CE-MS objects with known marker compounds (e.g., Paracetamol EOF marker) and need to extract their migration time positions to establish a calibration reference.

Use when immediately after importing raw LC-MS peak tables (e.g., Progenesis format) and before applying group or replicability filters.

Use when when you have a metabolomics abundance table with missing values and need to decide which imputation method to apply, or when designing a simulation to evaluate imputation performance.

Use when after converting MS/MS spectra to fixed-length vector representations using a pre-trained Word2Vec model (as in Spec2Vec), filter spectra before computing similarity scores to flag those where a large fraction of the observed intensity comes from peaks or neutral losses not present in the.

Use when you have a complete metabolomics data matrix (simulated or real abundance table) and need to create reproducible, controlled MNAR scenarios for evaluating imputation algorithm performance.

Use when apply fillPeaks after retention time alignment (whether XCMS or ncGTW) when feature matrices contain missing peaks across samples due to alignment gaps or detection failures.

Use when after sample alignment and feature grouping in untargeted LC-MS workflows, when the aligned feature table contains missing intensity values (NA or zero entries) due to features falling below the detection limit in some samples but being present above-threshold in others.

Use when when preparing raw HPLC column parameter arrays for featurization into feature vectors for retention time prediction models. Specifically apply this skill when column metadata contains empty strings (indicating missing diameter or pH values) or non-standard string encodings (e.g., '2.

Use when after feature extraction and quality control filtering (blank masking, sample dropping, normalization) have been applied, but before statistical analysis or machine learning.

Use when after feature alignment across multiple LC-MS/MS runs, when the unified feature list contains zeros or nulls for specific feature–sample pairs because peaks were not detected in those individual runs, but the feature was detected in other samples in the cohort.

Use when after mark_nas() has replaced non-NA missing-value codes (e.g., 0, 1) with R's NA in the exprs matrix of a MetaboSet object, and you need to decide whether to apply random forest imputation, simple imputation strategies, or defer imputation.

Use when after loading a feature table into memory when the table contains zero or missing values that represent true signal loss (not genuine absence), and you need to impute them before normalization, batch correction, or statistical analysis.

Use when you have metabolomics data (targeted LC/MS or untargeted GC/MS) with left-censored missing values below the limit of quantification (LOQ) or limit of detection (LOD), and you need to impute these values while preserving the underlying distributional structure and avoiding bias from.

Use when you need to measure how much a specific model capability or architectural feature contributes to prediction performance, especially when that capability is non-obvious or orthogonal to baseline methods.

Use when after a deep neural network model has completed training on LC-MS spectral peak classification data and you need to preserve the learned weights and architecture for downstream inference, validation on held-out test sets, or sharing with collaborators.

Use when when you have multiple candidate spectrum prediction models (e.g., FFN vs. GNN encoders, NEIMS vs. MassFormer vs. ICEBERG) and need to determine which performs better on a shared task like tandem mass spectrum prediction.

Use when when you have predictions from multiple independently trained models (e.g., ROASMI_1–ROASMI_5) for the same set of compounds in a reversed-phase liquid chromatography system at eluent pH ~2.7, and you need to estimate prediction reliability without ground-truth labels.

Use when after training a deep-learning model on paired MS/MS spectra with annotated structural similarity labels, use this skill to assess whether predicted similarity scores correlate with ground-truth reference similarities on data the model has never seen.

Use when you have a ProForma 2.0 peptidoform string (e.g., DLTDYLM[Oxidation]K) and need to extract the underlying peptide sequence and map modification positions to enable fragment ion annotation, mass calculation, or spectral matching.

Use when you have a pair of MS/MS spectra—one from a known compound and one from a structurally related modified (unknown) compound—and need to identify which atom(s) in the structure carry the modification.

Use when you need to simulate LC-MS/MS spectra for a specific biomolecule type (peptides, modified nucleosides, or other metabolites) and must choose which fragmentation model governs how parent ions break into fragment ions.

Use when after MS-DIAL peak character estimation has grouped LC-MS features into preliminary clusters based on peak shape and chromatographic similarity, and you need to select a single representative parental feature from each cluster to reduce redundancy before MS-FINDER annotation.

Use when evaluating whether a mass spectrometry data analysis platform (such as mzmine) provides complete module coverage across all advertised separation and ionisation techniques.

Use when when you need to understand how a multi-instrument mass spectrometry platform (like mzmine) decides which processing module receives a given dataset based on its declared data type (LC vs. GC vs. IMS vs. MS imaging).

Use when after a trained CNN model has generated molecular embeddings for query spectra, and you need to retrieve the most likely candidate molecules from a reference database.

Use when you have SMILES strings or 2D molecular structures of N-Me derived unsaturated sterol lipids (or other C=C-containing molecules) and need to generate 3D conformational and electronic structure data as input to a machine-learning CCS prediction model.