What problem does it solve?
This Skill prepares heterogeneous multi-omics datasets for joint analysis by normalizing per-assay measurements, correcting cross-platform batch effects, aligning feature identifiers, and handling missing values so downstream integration methods can operate reliably.
Core Features & Use Cases
- Assay-specific normalization: VST/TMM for RNA-seq, log2 + median centering for proteomics, M-value transform for methylation, and recommended methods for metabolomics.
- Cross-omics batch correction: Stack common-sample matrices and apply ComBat or model-based correction while preserving biological covariates.
- Feature alignment and mapping: Map protein or probe identifiers to gene symbols and aggregate to gene-level when needed.
- Missing value handling: Assess missingness, filter high-missing features, and impute (e.g., MinProb) per-assay where appropriate.
- Export & interoperability: Return a MultiAssayExperiment object or harmonized matrices (CSV/RDS) for MOFA2, mixOmics, or other integration tools.
- Use Case: Combine RNA-seq, proteomics, and methylation data from multiple cohorts, remove batch artifacts across platforms, map proteins to genes, impute proteomics gaps, and export a harmonized dataset ready for latent-factor integration.
Quick Start
Prepare my RNA-seq, proteomics, and methylation datasets for integration by normalizing each assay, correcting batch effects across platforms, aligning features to gene symbols, imputing missing values, scaling, and returning a harmonized MultiAssayExperiment object.
