Claude Skills by HolobiomicsLab

Use when when processing multiple PSM files from search engines (e.g., MaxQuant, MSGFPlus, Mascot) that omit fixed modifications from their output, or when PSM and spectrum files use inconsistent ID schemes, decoy naming conventions, or modification notations.

Use when when you need to expand or contract a mass spectrometry dataset by adding or removing allowed instrument types (e.

Use when when setting up NPLinker for natural products data mining and you have a TOML configuration file (nplinker.toml) that specifies root_dir, mode (local or podp), and paths to GNPS, AntiSMASH, and MIBiG directories.

Use when you have a molecular geometry (XYZ format) and need to explore its conformational space prior to quantum mechanical property calculation or when QCxMS2 workflow requires an ensemble of low-energy structures to enable systematic fragmentation pathways for EI mass spectrum simulation.

Use when after training or evaluating a classification model (e.g., MS2DeepScore or other neural networks) when you have paired arrays of predicted class labels and ground-truth labels and need to assess per-class prediction accuracy, false positive/negative rates, or class imbalance effects.

Use when you have a feature list from LC-MS preprocessing (e.g., asari output) and have already identified all pairwise feature matches to isotope and adduct patterns. Apply this skill when you need to separate feature matches into disjoint empirical compounds—i.

Use when you have spectral features annotated by both in silico structural prediction tools (e.g., SIRIUS/CANOPUS) and spectral library matching (GNPS), and you need a single authoritative chemical classification per feature rather than a list of competing candidates.

Use when you have detected a single chromatographic peak in DDA LC-MS/MS data that generated multiple MS2 fragmentation spectra (e.

Use when when you have structural annotations from multiple sources (in silico predictions via SIRIUS/CANOPUS and GNPS spectral library matches) for the same molecular features and need a single authoritative taxonomy assignment.

Use when you have a generic genome-scale metabolic model (SBML format) and cross-sectional omics data (RNA-seq, intracellular metabolomics, extracellular flux measurements from bioanalyzer or similar) from multiple biological samples (cell lines, conditions).

Use when when you have constraint-based metabolic models for multiple samples or cell lines, and you want to determine whether integrating multiple omics constraint types (nutrient availability, extracellular metabolite ratios, gene expression) produces distinct and biologically meaningful.

Use when you have constraint-based metabolic models with integrated multi-omics constraints (transcriptomics via Reaction Activity Scores, extracellular flux ratios via YSI bioanalyzer or LC-MS, nutrient availability bounds), and you need to determine whether differential enzyme expression.

Use when your Nextflow metabolomics workflow has been validated with Docker locally, but you need to deploy it on an HPC cluster that mandates Singularity containerization (e.g., Red Hat Enterprise Linux 8.

Use when your LC-HRMS metabolomics data (.mzML or .abf files) must be processed reproducibly across multiple machines (local workstations, HPCs, cloud) without manual tool installation, or when you need to enforce identical computational environments for peer review and long-term archival.

Use when when deploying a Nextflow workflow across multiple execution environments (local machines, HPC clusters) where tool versions, dependencies, or OS configurations may differ. Choose this skill specifically when you need to process LC-HRMS .mzML or .

Use when you have paired MS/MS spectra and molecular structures (SMILES or SDF format) and need to perform compound identification by retrieving the correct structure for an unknown spectrum.

Use when training embeddings from MS/MS spectra and you need to simultaneously enforce: (1) discrimination between spectra with different structural properties via contrastive learning on peak information and metadata embeddings, and (2) accurate reconstruction of embeddings from peak features via.

Use when you have a transformer encoder producing representations of tandem mass spectra and need to train it using contrastive learning with pairs of original and randomly masked spectra.

Use when when you need to understand how a multi-instrument mass spectrometry platform (such as mzmine) selectively routes data to different processing pipelines based on declared input type (LC, GC, IMS, or MS Imaging).

Use when you have collected or inherited sample-information metadata from multiple sources (e.

Use when when you have raw LC-MS/MS data in MGF format and need to prepare it for unsupervised topic modeling of metabolomics fragmentation patterns.

Use when when you have raw LC-MS-MS fragmentation spectra in MGF format and need to convert them into a corpus JSON file before running topic modeling.

Use when when deploying a Word2Vec-based spectral similarity model (such as Spec2Vec) on a new mass spectrometry dataset and needing to assess whether the pre-trained model's learned peak embeddings sufficiently represent the peaks in your query spectra.

Use when after identifying structural clusters (isotopologue groups, adduct groups, and cross-assay links) and assigning features to correlation clusters via hierarchical clustering.

Use when you have an aligned LCMS feature table (output from Eclipse or equivalent alignment tool) and need to identify candidate redundant features for clustering. It is most useful when features are suspected to originate from the same metabolite (e.

Use when after obtaining per-sample model predictions and metabolite signal intensities from a trained deep learning model (e.g., DeepMSProfiler) on LC-MS data from multiple disease groups, and you need to identify and visualize metabolite–disease associations as correlation strengths.

Use when after computing a correlation matrix (e.g., Pearson correlation across samples) on statistically significant LC-MS features, particularly when you need to inspect hierarchical dendrogram structure, validate cluster assignments from different flattening methods (constant-threshold vs..

Use when when you have preprocessed mass spectra (peak-filtered, metadata-cleaned) in supported formats (mzML, mzXML, msp, MGF, JSON) and need to compare all or many pairs of spectra to identify similar compounds, search spectral libraries, or build a similarity network for spectral clustering or.

Use when when you have imported and filtered mass spectrometry spectral data (from mzML, mzXML, msp, MGF, or JSON formats) and need to identify similar or related spectra within a dataset or against a reference library.

Use when you have a collection of deconvolved mass spectra (in MGF or mzTab format) from GC-MS analysis and need to group them into a molecular network to identify structural relationships and enable compound annotation.

Use when after normalization (Step 7) is complete and you have a clean feature matrix ready for multivariate statistical analysis.

Use when you have two sets of lipid annotations—one from baseline spectral library matching and one from an enhanced method (e.

Use when after identifying statistically significant features within individual LC-MS assays (e.g., via MB-VIP and permutation testing), use this skill when you have multiple parallel assays acquired in complementary ionization modes (e.

Use when when you have structural clusters from multiple LC-MS assays (e.g., positive and negative ion modes, or reversed-phase and HILIC methods) and need to identify which features across assays represent the same underlying metabolite.

Use when when you have antiSMASH-predicted BGCs and wish to link them to metabolomic data via structure prediction, but only BGCs with sufficient structural homology to characterized reference clusters will yield reliable predictions.

Use when you have two or more nontargeted LCMS feature tables from the same analytical method (same column, ionization mode, and acquisition parameters) and need to identify which features in one dataset correspond to features in another.

Use when you have a pre-trained MS/MS spectral embedding model evaluated on one or more source datasets (GNPS, MoNA, MTBLS1572) and need to verify that it performs well on independent, high-quality curated spectral libraries to claim robustness.

Use when when you have applied multiple scoring functions (e.g., strain correlation and IOKR) to rank genomic-metabolomic (GCF-MF or BGC-spectrum) links and need to verify that: (1) standardisation produces zero mean and unit variance across all links;

Use when when you have submitted the same MS/MS spectrum query to multiple domain-specific MASST tools and need to compare matches, combine ranked results, or generate cross-domain summary statistics.

Use when you have mass spectrometry data (LC–MS/MS, ion mobility, DIA) converted to MZA format and need to read or analyze it in multiple programming languages (Python and R), or share datasets with collaborators using different environments without requiring proprietary vendor libraries or format.

Use when your R-based Spectra analysis workflow requires a specific mass spectrometry algorithm (e.g., CosineGreedy similarity scoring, spectral normalization, or advanced filtering) that is available only in a Python MS package (matchms, spectrum_utils) and not yet in native R.

Use when you have a mature R analysis pipeline (e.g., using Spectra objects for MS data) but need access to specialized algorithms or calculations available only in Python libraries (e.g., CosineGreedy similarity from matchms).

Use when you have a pretrained RT-Transformer model checkpoint from a large, well-characterized chromatographic dataset (e.g., SMRT) and need to predict retention times for a different chromatographic method or instrument condition represented in a smaller, domain-specific dataset (e.g., PredRet).

Use when when you have predicted retention times from a DNN model trained on one chromatographic method (CM) and need to rank or filter metabolite candidates on a different chromatographic method, but have only a small set of calibration molecules (≥10) with known RTs on both source and target.

Use when when you have paired genomics (AntiSMASH BGC annotations) and metabolomics (GNPS spectra and molecular families) data from the same microbial strains and need to identify which biosynthetic pathways produce which observed natural products.

Use when you have already generated per-sample MS2 fingerprints (count vectors of MS2 peaks and neutral losses to the precursor) from matchms-processed spectra using spec2vec document representations, and you need to align these fingerprints into a single matrix to enable comparison across samples.

Use when you have run mass detection and chromatogram building independently on each LC-MS/MS sample and produced per-sample feature lists with m/z, retention time, and intensity values.

Use when you have detected feature tables from two or more LC-IMS-MS/MS samples and need to establish correspondence between features across samples to enable quantitative comparison, statistical analysis, or consensus feature calling.

Use when after feature extraction (MS1 peak picking, MS2 recognition, or targeted list extraction) from multiple individual samples and before annotation.

Use when you have collected feature lists in CSV format from two or more MS acquisition methods (e.g., LC-MS vs. LC-IMS-MS), processing software packages (e.g., vendor-specific vs. open-source), or instrument platforms (e.