Metabolomics
LC-MS, GC-MS, and NMR data analysis — QC, preprocessing, multivariate statistics, metabolite annotation, and pathway enrichment
lobster-metabolomics
FreeIntermediate
Metabolomics data analysis: QC, preprocessing, multivariate statistics (PCA/PLS-DA/OPLS-DA), metabolite annotation, and pathway enrichment
Input
LC-MSGC-MSNMRCSVH5AD
Output
QC ReportPCA/PLS-DA PlotsAnnotated MetabolitesPathway EnrichmentLipid Classes
Agents (1)
└── metabolomics_expert — Full metabolomics workflow orchestration
pip install lobster-metabolomics
Agents
metabolomics_expert
The metabolomics expert handles end-to-end metabolomics analysis across LC-MS, GC-MS, and NMR platforms. It detects the input platform automatically and applies appropriate preprocessing, statistical, and annotation workflows.
Capabilities:
- Platform detection (LC-MS, GC-MS, NMR) and format-aware loading
- Quality assessment and outlier detection
- Preprocessing: filtering, imputation, normalization (PQN, TIC, median)
- Batch correction
- Univariate statistics (t-test, ANOVA, fold change)
- Multivariate statistics (PCA, PLS-DA, OPLS-DA)
- m/z annotation and metabolite identification
- Lipid class analysis
- KEGG pathway enrichment
Example Workflows
LC-MS QC and Preprocessing
User: Load my LC-MS metabolomics data and run quality control
[metabolomics_expert]
- Detects LC-MS platform from feature columns
- Runs QC: coefficient of variation, missing value analysis, outlier detection
- Applies TIC normalization and KNN imputation
- Generates QC summary with distribution plotsMultivariate Statistical Analysis
User: Run PCA and PLS-DA comparing treatment vs control groups
[metabolomics_expert]
- Performs PCA for unsupervised overview
- Fits PLS-DA model with cross-validation (Q2, R2 metrics)
- Identifies VIP > 1 features (discriminating metabolites)
- Generates score plots and loading plotsMetabolite Annotation and Pathway Analysis
User: Annotate the top differentially abundant features and run pathway enrichment
[metabolomics_expert]
- Matches m/z values against reference databases
- Assigns putative metabolite identities with confidence levels
- Groups lipid species by class (if applicable)
- Runs KEGG pathway enrichment on annotated metabolites
- Returns enriched pathways with p-values and metabolite mappingsDependencies
| Library | Purpose |
|---|---|
| anndata | Data container for metabolomics matrices |
| numpy | Numerical computation |
| pandas | Tabular data manipulation |
| scipy | Statistical tests and signal processing |
| scikit-learn | PCA, PLS-DA, OPLS-DA, preprocessing |
| statsmodels | Univariate statistics and multiple testing correction |
These are installed automatically with the package.
Services
lobster-metabolomics includes 4 domain-specific services:
| Service | Purpose |
|---|---|
| MetabolomicsQualityService | QC metrics, outlier detection, CV analysis |
| MetabolomicsPreprocessingService | Filtering, imputation, normalization, batch correction |
| MetabolomicsAnalysisService | PCA, PLS-DA, OPLS-DA, univariate statistics |
| MetabolomicsAnnotationService | m/z annotation, lipid classification, pathway enrichment |
Configuration
# .lobster_workspace/config.toml
enabled = ["metabolomics_expert"]