mixOmics

mixOmics offers multivariate methods for the exploration, integration, and variable selection of biological datasets where the number of variables (e.g. genes, proteins, metabolites) exceeds the number of samples (patients, cells, mice).

Key Features:

• Dimension Reduction: Reduces data dimension using instrumental variables (components) defined as combinations of all variables.
• Graphical Outputs: Produces graphical outputs to visualize relationships and correlation structures between integrated datasets.
• Variable Selection: Implements sparse multivariate models to identify key variables that are highly correlated and/or explain biological outcomes of interest.
• Missing Value Handling: Handles missing values without requiring deletion of entire rows with missing data.
• N-integration: Combines multiple datasets using variants of Generalised Canonical Correlation Analysis.
• P-integration: Combines multiple datasets using variants of multi-group Partial Least Squares.

Scientific Applications:

• Omics Data Integration: Analyzes high throughput sequencing data including transcriptomics, metabolomics, proteomics, and metagenomics.
• Beyond Omics: Analyzes non-omics data types such as spectral imaging.
• Correlation Analysis: Identifies highly correlated variables across integrated biological datasets.
• Outcome Prediction: Identifies variables that explain biological outcomes of interest.

Methodology:

The methods implemented include variants of generalised Canonical Correlation Analysis, sparse Partial Least Squares, sparse Discriminant Analysis, N-integration with variants of Generalised Canonical Correlation Analysis, and P-integration with variants of multi-group Partial Least Squares.