Claude Skills by HolobiomicsLab

Use when you have raw molecular datasets (e.g., METLIN-CCS, CCSBase) with SMILES strings, 3D coordinates, adduct information, and ground-truth collision cross section labels, and you need to format them for GNN training and held-out test evaluation with consistent standardization.

Use when you have raw molecular structures in SMILES or SDF format and need to prepare them as input for BitterPredict.m or similar descriptor-based classifiers.

Use when when you have validated SMILES strings or canonical molecule objects from RDKit and need to convert them into the fixed-size numerical tensor format expected by a deep learning model (e.g., PS2MS).

Use when you have a collection of chemical structures (SMILES, InChI, SDF, or mol formats) and need to train or apply a machine learning model for retention time prediction or molecular property estimation.

Use when after SMILES standardization when you have a table of translated SMILES strings (e.g., interim/tables/1_translated/structure/smiles.tsv.

Use when you have untargeted metabolomics peak intensity data and spectral groupings (Molecular Families or Mass2Motifs) but lack confident chemical annotations or want to avoid pathway database dependency.

Use when you have loaded a collection of molecular fingerprint vectors (e.g., from biosynfoni fingerprints deposited in Zenodo) and need to assess their statistical properties before using them for classification, similarity search, or method validation.

Use when you have natural product molecules (or compounds from natural product-like databases such as COCONUT or ZINC) in structural format (SMILES, InChI, or SDF file) and need a fingerprint representation optimized for biosynthetic-class prediction, structural clustering, or bioinformatic feature.

Use when you have annotated metabolite structures (with SMILES strings) from a reference library (e.

Use when you have received a JSON response from the CSI:FingerID web service endpoint after submitting a fragmentation tree or tandem mass spectrum query, and you need to extract the predicted molecular fingerprint representation and associated scoring metrics for compound identification or CANOPUS.

Use when when evaluating how well mass spectral similarity scores correlate with actual chemical structure for annotated spectral pairs (e.g., spectra with InChIKey metadata).

Use when you have a calibrated FT-ICR transient (ESI_NEG or similar ionization mode) and need to annotate each detected m/z peak with its most likely elemental composition.

Use when when you need to constrain a large metabolite database to a specific instrumental range (e.g., m/z 100–1000) before generating virtual chemical mixtures for LC-MS/MS simulation, or when you need to verify that a reported filtered database count can be reproduced from raw database files.

Use when you have MS/MS fragmentation spectra (from Orbitrap or Q-TOF instruments) in MGF format with known precursor m/z, adduct type, and collision energy, and you need to generate ranked molecular formula candidates.

Use when when you have a feature list from HRMS with tentatively assigned molecular formulas (from in silico tools or databases) and need to assess formula plausibility before applying downstream PFAS-specific filters.

Use when when you have encoded spectral features (from a CNN featurizer applied to 1D 1H and/or 13C NMR spectra) and a set of candidate molecular fragments predicted for a molecule with ≤19 heavy atoms, and you need to determine the correct connectivity and assembly order to recover the.

Use when you have raw SMILES strings from a chemical database (e.g., CCSBase, METLIN, or custom sources) and need to feed them into a graph neural network model.

Use when you have molecular identifiers (SMILES strings or molecular structure files) that need to be converted into node-edge graph tensors for input to message passing neural network models like chemprop or chemprop-IR.

Use when you have a GNPS mass spectral molecular network (classical or feature-based) and MS2LDA-derived Mass2Motif data, and you need to annotate network nodes with both chemical class information from GNPS library matches and substructural motifs from MS2LDA to enable joint interpretation of.

Use when you have a GNPS mass spectral molecular network (in .graphml or Cytoscape format) and wish to annotate its nodes and edges with chemical class assignments from the GNPS library and/or MS2LDA motif probabilities from an independent LDA experiment.

Use when after generating candidate transformed structures from biotransformation rules and when you have MS/MS spectral feature data that you wish to organize into putative molecular families.

Use when you have untargeted metabolomics data (e.g., LC-MS/MS spectra) and need to organize compounds by structural relatedness to enable structure discovery for unknown metabolites.

Use when you have a GNPS-generated classical or feature-based molecular network (in graphml or JSON format) and corresponding MS2LDA or chemical class assignment data, and you need to embed substructural motif identifiers, confidence scores, or chemical class labels as node/edge attributes for.

Use when after GNPS_GC molecular networking job completion, when you have retrieved raw network output files and need to extract, validate, and structure the network topology for further metabolite assignment, comparative network analysis, or visualization.

Use when you have LC-MS/MS DDA data from one or more samples and need to organize fragmentation spectra by similarity relationships to support compound annotation, enable cross-sample comparisons, and identify known and unknown metabolites sharing structural features.

Use when when you have molecular structures (SMILES or graph formats) and need to predict a physicochemical or spectral property (e.g., infrared spectra) using a graph neural network architecture. Use this skill when the base model architecture (e.

Use when you have a trained GNN model predicting CCS values from molecular graphs and need to understand which structural features (node and edge attributes) are most influential for specific predictions or across a test set.

Use when after an end-to-end neural model (CNN + transformer) has generated predicted molecular structures (formula and connectivity) from 1D NMR spectra.

Use when you are receiving molecular structures from external sources (COCONUT database, ZINC database, user-provided chemical data) in varying formats (SMILES strings, InChI identifiers, SDF files), and you need to unify them into a single canonical representation before computing biosynfoni.

Use when you have trained a multitask NMR-to-structure model and need to quantify its predictive accuracy on held-out test molecules.

Use when your goal is to assess whether a pretrained NMR2Struct model trained on molecules ≤19 heavy atoms can generalize to larger, more complex molecules, or whether accuracy degrades significantly.

Use when when you have metabolite structures (as SMILES strings) and need to predict their observable m/z ions under non-standard ionization conditions imposed by a derivatizing matrix reagent (such as FMP-10).

Use when when a trained Siamese neural network model makes predictions on new spectrum pairs and you need to identify and exclude high-uncertainty predictions to improve RMSE.

Use when after completing the MS2LDA LDA modeling step when you have a JSON-serialized inferred motifset (Mass2Motifs with fragment and neutral-loss patterns) and need to annotate those motifs by comparing them against a curated MotifDB reference database to identify known structural subpatterns.

Use when after LDA inference has produced a trained motifset (motifset.json or motifset_optimized.json) with Mass2Motif probability distributions over fragments and neutral losses.

Use when after executing MassQL queries against a MotifDB reference database and retrieving ranked motif matches, when you need to determine which database entries represent true structural correspondence versus spurious matches, and to decide whether a motif's top-ranking hit is sufficiently.

Use when you have a GNPS-generated classical or feature-based mass spectral molecular network (graphml or JSON format) and a corresponding MS2LDA experiment with Mass2Motif assignments on the same spectra, and you want to visualize and quantify which structural motifs are shared within and across.

Use when you have a preprocessed peak table from tandem MS/MS data (e.

Use when you have MS data in a new format or storage system (e.g., a custom database, HDF5 file, or proprietary raw file) and need to make it accessible to Spectra-based analysis workflows without forking the Spectra package itself.

Use when implementing or validating a new MsBackend class that stores m/z and intensity values, or when assigning peak data to an existing backend. Triggers include: (1) implementing a replacement method (e.g., mz<- or intensity<-) for a backend class;

Use when you have performed lipid identification in MS-DIAL and need to pass the results to LipoCLEAN or another downstream quality-filtering tool. The skill is required whenever you are preparing MS-DIAL output for consumption by external analysis pipelines that expect standardized export formats.

Use when when you have raw LC-HRMS metabolomics data in .mzML or .abf format and need to perform untargeted feature detection with chromatographic alignment across multiple samples.

Use when you are preparing to run LipoCLEAN on MS-DIAL output and need to create or update a configuration file, or you have switched between MS-DIAL 4 and MS-DIAL 5 data and need to verify that your options file and trained model are compatible with the current version's column naming and scaling.

Use when when evaluating or designing a mass spectrometry data analysis platform, and you need to verify that every supported separation/ionisation technique (LC, GC, IMS, MS Imaging) is covered by at least one processing module.

Use when you have millions of MS/MS spectra in mzML, mzXML, or MGF format that have been converted to low-dimensional vectors via feature hashing, and you need to identify which spectra are similar enough to cluster together.

Use when you have raw or preprocessed MS/MS spectra in one of the supported formats (MGF, mzML, mzXML, JSON, MSP, mzXML, pickled matchms objects, or USI) and need to extract peak lists (m/z and intensity pairs) along with metadata (precursor m/z, charge, ionization mode) to feed into MS2Query or.

Use when you have an experimental MS/MS spectrum (e.g., from MassBank or local data) and need to identify significant fragment ions above noise, assign occurrence scores to peaks, and generate a CSV-formatted library entry for use in metabolite feature annotation pipelines.

Use when you have raw MS/MS spectra (in MGF or mzML format) with unscaled peak intensities and noise artifacts, and you plan to rank chemical formulas, predict adducts, or score precursor–spectrum agreement using a machine learning model such as MIST-CF.

Use when immediately after loading a raw GC-MS CSV file and before executing the spreadOut() function. Use it when you have received peak table data from an instrument vendor (e.

Use when you have raw MS intensity data paired with known-concentration calibration standard measurements, and you need to convert intensities to absolute or relative concentrations.