Fig. 2: Characterization of multi-omics molecular landscape for osteoporosis.
From: Multi-modal molecular determinants of clinically relevant osteoporosis subtypes
a Sample group distribution in the PLSDA space. b Percentages of discriminative features at different omics data levels. c Percentages of phenotype-associated features at different omics data levels. d Co-expression network of features for different sample groups at corresponding omics data level. e Co-expression network modules of features and their phenotype associations for different sample groups at particular omics data level. f Shared information between different omics datasets estimated by RV index. g The multi-omics unsupervised learning for osteoporosis multi-omics atlas by our deep latent space fusion model based on joint auto-encoder and self-expression technologies.
