Co-presence multiomic network graphs for predicting barley phenotypes

Overview

This research proposes a computational method that uses co-presence multiomic network analysis to link barley phenotypic traits—including leaf angle, heading time, and plant height—with specific cell types and cellular functions. By integrating transcriptomic, genomic, and metabolomic datasets, the approach addresses a key challenge in bulk RNA-sequencing: heterogeneous cell populations that can obscure biological interpretation. The resulting network graphs enable more accurate prediction of crop phenotypes, supporting improved breeding decisions and digital twin modeling for agricultural applications.

Technical specifications

Methodology:

• Constructs co-presence network graphs combining transcriptomic, genomic (single nucleotide polymorphism), and metabolomic data from barley samples
• Clusters network components using the Markov Cluster Algorithm to identify groups of co-inherited transcripts, gene variants, and metabolites
• Performs cellular composition deconvolution of bulk RNA-sequencing samples using EcoTyper to account for heterogeneous cell populations
• Evaluates clusters for their ability to differentiate measured phenotypes and for enrichment in known gene pathways
• Validates predictive power by applying clusters to independent datasets for phenotype prediction

Key features:

• Multiomic integration of transcriptomics, genomics, and metabolomics in a single network framework
• Cell-type-aware analysis that accounts for tissue heterogeneity in bulk sequencing data
• Pathway enrichment analysis linking co-inheritance patterns to biological function
• Applicability to digital twin modeling through graph neural networks and generative adversarial networks

Technology readiness level

The methodology builds on previously validated approaches, including co-expression transcriptomic network analysis for cell type characterization, network-based identification of co-inherited variants, and co-abundance network analysis for metagenomic data. The researcher has demonstrated the ability to combine metabolomic and transcriptomic data for co-presence analysis. The current project extends these capabilities to barley multiomic data, with validation planned against published datasets. The approach is at an early-to-mid research stage, with computational methods established but barley-specific applications still undergoing validation.