Linking genotype to phenotype in multi-omics data of small sample.

Area of Science:

• Genomics
• Systems Biology
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) link genotype to phenotype but offer limited mechanistic insights.
• Integrating multi-omics data improves genotype-phenotype prediction accuracy.
• Existing multi-omics integration methods overlook intra- and inter-omics associations and require large, single sample sets.

Purpose of the Study:

• To develop a genotype-phenotype association method utilizing multi-omics data from small sample sizes.
• To overcome limitations of current methods that ignore omics associations and demand large sample cohorts.
• To enhance the accuracy of predicting biological associations between genotype and phenotype.

Main Methods:

• Gene clustering via protein-protein interaction networks and gene expression data.
• Group lasso for screening gene clusters and expression quantitative trait locus (eQTL) data for corresponding SNP clusters.
• Construction of three-layer network blocks integrating SNP clusters, gene clusters, and phenotypes for analysis.

Main Results:

• The proposed method effectively handles multi-omics data from small sample sets.
• The method demonstrates superior performance compared to existing approaches in two independent datasets.
• The approach provides a robust framework for genotype-phenotype association predictions.

Conclusions:

• The developed method successfully addresses the challenge of genotype-phenotype prediction with limited multi-omics data.
• This approach offers a valuable resource for future research involving the integration of diverse omics data.
• The findings support the utility of network-based integration for understanding genotype-phenotype relationships.