This study introduces and illustrates the potential of an integrated multi-omics approach in investigating the underlying biology of complex traits such as childhood aggressive behavior. Using multivariate statistical methods, we integrated 45 polygenic scores (PGSs) based on genome-wide SNP data, 78,772 CpGs, and 90 metabolites for 645 twins (cases=42.0%, controls=58.0%). The single-omics models selected 31 PGSs, 1614 CpGs, and 90 metabolites, and the multi-omics biomarker panel comprised 44 PGSs, 746 CpGs, and 90 metabolites. The predictive accuracy in the test (N=277, cases=42.2%, controls=57.8%) and validation data (N=142 participants from a clinical cohort, cases=45.1%, controls=54.9%) ranged from 43.0% to 57.0% for the single- and multi-omics models. The average correlations across omics layers of omics traits selected for aggression in single-omics models ranged from 0.18 to 0.28. In the multi-omics model higher correlations were found and we describe five sets of correlational patterns with high absolute correlations (|r| ≥ 0.60) of aggression-related omics traits selected into the multi-omics model, providing novel biological insights.