Multiblock partial least squares (MB-PLS) is an extension of the PLS method that performs N-integration on multiple datasets measured across the same samples, using a design matrix to control the relationships between them. Like two-omics PLS, MB-PLS can be supervised (regression) or unsupervised depending on the chosen mode. In sparse MB-PLS (MB-sPLS), lasso penalisation is applied on the loading vectors to identify the most important variables.
Data used on this page:breast.TGCA
Key functions used on this page:block.pls()block.spls()plotLoadings()plotIndiv()plotVar()
Related case studies:
Multiblock sPLS Gastrulation Case Study
References:
1. Tenenhaus A and Tenenhaus M. Regularized generalized canonical correlation analysis. Psychometrika, 76(2):257–284, 2011.