MoNETA: MultiOmics Network Embedding for SubType Analysis

Abstract

Cells are complex systems whose behavior emerges from a huge number of reactions taking place within and among different molecular districts. The availability of bulk and single-cell omics data fueled the creation of multi-omics systems biology models capturing the dynamics within and between omics layers. Powerful modeling strategies are needed to cope with the increased amount of data to be interrogated and the relative research questions. Here, we present MultiOmics Network Embedding for SubType Analysis (MoNETA) for fast and scalable identification of relevant multi-omics relationships between biological entities at the bulk and single-cells level. We apply MoNETA to show how glioma subtypes previously described naturally emerge with our approach. We also show how MoNETA can be used to identify cell types in five multi-omic single-cell datasets.

Figure 1.

Representation of multi-omics data integration techniques. Fusion-based approaches unite diverse omics data into a singular model. Separation-based approaches build intermediate models for each omics, merging insights for a comprehensive analysis. Transformation-based approaches apply graph or kernel-dependent algorithms before integration, enhancing entity description.

Figure 2.

MoNETA workflow. MoNETA workflow for the analysis and visualization of multi-omics networks comprises four steps. The initial step involves data acquisition and pre-processing. The second step involves computing single omics networks to understand the underlying relationships between each omics data. This step is followed by multi-omics network integration, where individual networks are combined into a multiplex network. Random Walker is then applied to the multiplex network to identify nodes with similar attributes. The final step in the workflow involves embedding and dimensionality reduction of the data, which is necessary for visualizing high-dimensional data that highlights the underlying patterns and relationships in the data.

Figure 3.

Glioma sub-typing. Embedding of glioma data. Colors show glioma subtypes identified in (40) while shapes are associated with IDH sample status. (A) Single-omics embeddings. (B) MoNETA embeddings. (C) WNN embeddings. (D) MOFA embeddings. (E) Comparison of MOFA, WNN and MoNETA embeddings using accuracy and NMI scores.

Figure 4.

Single-cell LUNG-CITE, PBMC-TEA and paired clustering. Plots comparing the projections of cells obtained using data from individual omics layers with the MoNETA multi-omics integration for: (A) lung derived single-cell PBMC bi-modal assay; (B) blood-derived PBMC tri-modal assay; (C) adult mouse-brain derived hepta-modal single-cell Paired assay. Colors show cell types as identified in their annotation.