Claude Skills by HolobiomicsLab

Use when after filtering LC-MS features by statistical significance (e.g., p-value < 0.01) and you need to link individual m/z features into structural clusters representing the same metabolite in different ionization states or isotopic forms.

Use when you have loaded sqMass files containing pre-extracted transition group chromatograms and need to isolate chromatographic traces within a specific retention time interval—either defined by OpenSwath feature metadata (apex retention time ± margin) or by manual user selection.

Use when you have mzML/mzXML chromatogram files from Thermo, Waters, or Bruker instruments and need to extract MS1 and MS2 scans matching both a target m/z value AND a known or suspected retention time range.

Use when you have acquired LC-MS data and need to verify run quality before proceeding to metabolite identification or quantification. Retention time checking is essential when: (1) you have established expected retention time ranges for known internal standards or reference compounds;

Use when you have LC-MS/MS spectral data (in MGF, mzXML, mzML, or mzData format) and corresponding genomic sequence data (raw FASTA nucleotide sequences or genome mining tool outputs like antiSMASH .final.

Use when you have LC-MS/MS spectra (MGF, mzXML, mzML, or mzData format) and either raw genome nucleotide sequences or antiSMASH/BOA genome mining tool output, and you need to identify which RiPPs are present in your sample.

Use when when you have genomic sequences (assembled contigs or antiSMASH/BOA mining results) and want to match experimental tandem mass spectra against predicted RiPP structures.

Use when after identifying precursor peptides from genome assemblies via BGC mining, when you need to enumerate the chemical space of PTM variants (lantibiotic, lassopeptide, cyanobactin, or other RiPP classes) before matching tandem mass spectra against a constructed database using Dereplicator or.

Use when you have: (1) tandem MS/MS spectra in MGF, mzXML, mzML, or mzData format from LC-MS/MS analysis; (2) a set of predicted RiPP precursor peptides derived from genomic biosynthetic gene cluster mining (via antiSMASH, BOA, or raw FASTA);

Use when after drift correction of LC-MS peak intensity data, when you need to identify metabolic features with excessive internal spread (within-group variability in QC samples) or poor biological-to-technical reproducibility (QC-versus-sample spread).

Use when after training a NeatMS CNN model on labeled MS1 peaks and generating predictions on a held-out test set, compute ROC-AUC to assess whether the model achieves the target discrimination threshold (AUC ≥ 0.9) without evidence of overfitting.

Use when you have raw or converted mass spectrometry data (CE-MS or LC-MS in mzXML or mzML format) and need to identify candidate metabolite regions before feature extraction.

Use when after segmenting raw profile LC-MS data into candidate ROIs containing potential peaks, and before feeding ROI data to a CNN-Transformer peak detection network.

Use when when you have paired predictions and ground-truth structural similarity labels (e.g., predicted Tanimoto scores from a neural network and reference Tanimoto scores from RDKit Daylight fingerprints) and need to report a single scalar metric of model prediction error across all pairs.

Use when when evaluating a regression or similarity prediction model and you need to understand whether prediction error is uniform across the outcome space or concentrated in particular ranges (e.g., low structural similarity vs. high similarity scores).

Use when you have a trained regression model (e.g., a neural network or similar predictor) and a held-out test set with ground-truth continuous labels, and you need to measure whether the model's predictions match the true values.

Use when when you have implemented conditional routing logic in the GNPS_MASST codebase and need to verify that spectrum submissions with explicit domain-context selections (e.

Use when when preparing XCMS peak tables for quality classification and you observe that the default RSD threshold (0.3 or 30%) is either too permissive (retaining noisy EICs) or too stringent (discarding valid signals).

Use when you have centroided mzML or mzXML LC-MS files from a single batch run that exhibit systematic retention-time drift between samples, one or more designated QC reference file(s), and you need to harmonize RT coordinates across all samples before feature detection to reduce false positive and.

Use when your feature table includes QC (quality control) sample replicates distributed throughout the analytical run sequence, and you observe systematic intensity variation correlated with sample injection order or batch identifier—typical indicators of instrumental drift in untargeted LC-MS.

Use when when you have implemented or adopted a new clustering or analysis tool and need to validate that it meets stated runtime claims on a representative production-scale dataset. Particularly important when the tool uses hardware acceleration (GPU) and the claimed speedup is a core contribution;

Use when you are designing a new backend or data container that must integrate seamlessly with an existing Spectra-based workflow. You have identified a virtual parent class (e.

Use when you are extending the MsBackend virtual class to create a new backend for storing MS spectra data and need to define the internal data structure. Specifically: when you have multiple types of spectra variables (e.

Use when you have tab-delimited metabolomics data with columns for aliquot identifiers, compound names, peak areas (primary and internal standard), sample type (QC, study sample, calibration), batch labels, and injection times, and need to construct a single unified object for batch correction.

Use when after MZmine feature detection and molecular networking on a single LC-MS/MS DDA sample, when you have a feature table (with retention time, m/z, fragmentation spectra) and sample-level taxonomical metadata, and you want to assign both spectral identity and predicted chemical structure to.

Use when you have a raw peak table (CSV format, from any of 12 supported LC-MS software tools or standardized format) and a separate label file that assigns each sample to an experimental class (e.

Use when when aligning detected features across multiple LC-IMS-MS/MS samples and you need to identify which input samples contributed to each consensus feature cluster, especially to filter out spurious or low-confidence alignments, validate clustering completeness, or perform sample-specific.

Use when when you have loaded centroided .mzML files into a Spectra object and plan to use TARDIS (tardisPeaks) with an MsExperiment object rather than file paths, and you need TARDIS to distinguish QC runs from sample runs for separate quality metric calculation, polarity filtering, and.

Use when when you have an aligned MemoMatrix (sample-by-feature occurrence matrix where features are MS2 peaks and neutral losses) and corresponding sample annotations (especially blank/control sample labels), and you need to exclude background-derived peaks and losses before applying visualization.

Use when when you have LC-MS peak table data in Excel format (e.g., from MS-DIAL peak picking) with separate compartments for sample information, feature properties, and intensity values, and you need to construct a structured R object that enforces consistency between sample metadata (e.

Use when when you have retrieved a large, heterogeneous collection of tandem MS files from ReDU or MassIVE and need to isolate a subset sharing specific sample characteristics (e.

Use when when you have loaded (un)targeted metabolite data into a Metaboprep object and need to exclude samples with excessive missing values before quality control or statistical analysis. This is typically applied early in the QC pipeline when sample-level data completeness is a concern (e.

Use when you have a merged MGF file (e.g., from MZmine output) containing MS/MS spectra from multiple biological or environmental samples, and you need to process each sample independently through annotation pipelines (e.

Use when after generating aligned MS2 fingerprints (sample-by-fingerprint matrices) from metabolomics data when you need to visually inspect sample clustering, identify sample similarities, or detect batch effects and RT shifts across different LC methods or mass spectrometer technologies.

Use when you have a GNPS task ID from a completed molecular networking workflow (METABOLOMICS-SNETS, METABOLOMICS-SNETS-V2, FEATURE-BASED-MOLECULAR-NETWORKING on GNPS1, or classical_networking_workflow / feature_based_molecular_networking_workflow on GNPS2) and need to access the resulting spectral.

Use when when beginning mass alignment in a multi-sample LC-MS metabolomics study, before constructing the MassGrid.

Use when after mass tracks have been aligned across samples into a MassGrid structure and retention time calibration dictionaries (rt_cal_dict) have been computed for each sample during prior alignment steps.

Use when after feature detection but before statistical analysis, when your study includes blank samples (e.g., solvent or extraction blanks) and you want to remove features that fail to show meaningful enrichment in actual study samples relative to blank contamination.

Use when you have unaligned MS2 spectra from multiple metabolomics samples (in mzML, mzXML, or MGF format) and need to compare them without relying on retention time or aligned m/z features.

Use when after loading multiple LC-MS .mzML files into an MsExperiment object using MsBackendMzR backend, when you have a documented injection sequence (e.

Use when you have a new or modified LC-MS data processing tool and need to determine whether it can handle production-scale sample cohorts (50–100+ samples) on modest hardware (single-core CPU, ≤16 GB RAM).

Use when when you have generated a scan index from rawrr::readIndex() on a Thermo .

Use when you have a Thermo Orbitrap .raw file and need to (1) verify that a targeted acquisition method (e.g., PRM) maintains consistent scan spacing across all cycles; (2) extract only scans matching a specific precursor ion and fragmentation method;

Use when after performing LOWESS regression on landmark peak RT pairs between a sample and reference, you need to encode the learned RT transformation as a reusable, memory-efficient lookup table that can be applied during feature alignment without recomputing the regression for every sample.

Use when when processing LC-MS data with multiple overlapping m/z scan windows and observing sawtooth-pattern distortions in EICs during tardisPeaks() execution.

Use when when uploading or ingesting a new paired omics project JSON document into the Pairing Omics Data Platform, or when programmatically submitting projects via the OpenAPI interface. Apply this skill before persisting the document to disk or indexing it for search.

Use when a user uploads a JSON project file to the platform and you need to verify it matches the required format defined in app/public/schema.json before accepting it into the database.

Use when you have prediction arrays (model outputs) and ground-truth label arrays from a classification task and need to compute confusion matrices, accuracy scores, or other performance metrics for visualization or quantitative evaluation.

Use when after computing link scores (e.g., strain correlation, IOKR, or combined scores) across GCF-MF pairs, use this skill to assess whether scores achieve sufficient separation between validated links and the background population.

Use when when comparing raw link scores (strain correlation or IOKR) across different GCF-MF or BGC-spectrum pairs and you need to distinguish true positive links from background noise. Raw scores depend on overlap size and are incomparable across links with different cluster sizes;