mixOmics is primarily focused on dimension reduction and data exploration. Hence, it has a range of methods which allow for the graphical representation of omics data. Both the relationship between samples and variables can, and should, be explored. All graphical functions in the package are based on S3 methods, such that all methods are generalisable and can take almost any different type of object (eg. cca, spca or plsda) as input.
The graphical methods are split into two categories;
- Sample Plots allow for the visualisation of similarities between samples.
- Variable Plots aid in understanding the importance of the variables in defining the components, and visualise their relationships.
Sample Plots
plotIndiv() – Sample Plot: Displays samples as points projected onto selected latent components, with similar samples clustering together. Applicable to all mixOmics methods.
plotArrow() – Arrow Plot: Represents paired coordinates as arrows, where tip and tail indicate sample positions in different datasets. Short arrows suggest strong agreement, while long arrows indicate disagreement. Used for N-integration frameworks.
Variable Plots
plotLoadings() – Loadings Bar Plot: Displays loading vectors as a barplot, where each bar represents a variable’s contribution to a component. Suitable for most dimension reduction methods in mixOmics.
plotVar() – Correlation Circle Plot: Shows correlations between original variables and selected components, also revealing relationships between variables. Used in PCA, CCA, and PLS methods.
biplot() – Biplot: Similar to plotVar(), but also includes sample projections, helping to visualize how variables explain certain samples. Suitable for PCA and PLS.
cim() – Clustered Image Map (CIM): Generates color-coded heatmaps for high-dimensional data, useful for both single omics and two-dataset integration.
network() – Relevance Network Graph: Depicts associations between variables from two omics datasets, showing complex correlation structures and clustering. Used for two-dataset integration.
circosPlot() – Circos Plot: Extends relevance networks and CIMs to visualize associations across multiple datasets using a circular layout. Suitable for N-integration contexts.
