Prediction of plant complex traits via integration of multi-omics data

The formation of complex traits is the consequence of genotype and activities at multiple molecular levels. However, connecting genotypes and these activities to complex traits remains challenging. Here, we investigated whether integrating different omics data could improve trait prediction. We built prediction models using genomic, transcriptomic, and methylomic data from the Arabidopsis 1001 Genomes Project for six Arabidopsis traits, and found that transcriptome- and methylome-based models had performances comparable to those of genome-based models. However, when comparing models for flowering time prediction, we found that models built using different omics data identified different benchmark genes. Nine novel genes identified as important for flowering time from our models were experimentally validated as regulating flowering. In addition, we found that gene contributions to flowering time prediction are accession-dependent and that distinct genes contribute to trait prediction in different genetic backgrounds. Models integrating multi-omics data performed best and revealed known and novel gene interactions, extending knowledge about existing regulatory networks underlying flowering time determination. These results demonstrate the feasibility of revealing molecular mechanisms underlying complex traits through multi-omics data integration.