kimod

kimod integrates and analyzes mixed omics data using distance metrics, bootstrap resampling of residual matrices, and biplot projection to support interpretation of high-throughput genomics and molecular biology datasets.

Key Features:

• Distance Options: Provides distance metrics for numeric and categorical variables across multiple data tables to measure similarities and dissimilarities in mixed omics datasets.
• Bootstrap Resampling Techniques: Applies bootstrap resampling to residual matrices to estimate variability and construct confidence ellipses for projections of individuals and variables.
• Biplot Methodology: Uses biplot projection to map gene expression variables onto a compromise space for interpreting relationships between genes and samples.

Scientific Applications:

• Mixed omics integration: Integrates numeric and categorical omics tables for combined analysis of heterogeneous high-throughput genomics data.
• Uncertainty quantification: Quantifies variability in projections via bootstrap-derived confidence ellipses computed on residual matrices.
• Visualization of gene-sample relationships: Visualizes and interprets relationships between gene expression variables and samples through biplot projections in a compromise space.

Methodology:

Implemented in R within the Bioconductor project; methods explicitly include distance metrics for numeric and categorical variables, bootstrap resampling on residual matrices to derive confidence ellipses, and biplot projection of gene expression variables onto a compromise space.