Outcomes and traits in mental disorder research are complex and attributable to molecular changes at multiple levels. The analysis of mental disorder omics data is extremely challenging, with the complexity of outcomes, weak signals, high dimensionality, and unknown regulations among omics changes. To tackle such challenges, we propose taking an integrative perspective: the integrative analysis of multiple independent datasets can effectively increase power and lead to more reliable estimation; and the integrative analysis of multiple types of omics data can lead to a better understanding of mental disorder biology and clinically more plausible models. For such analysis, we have developed a series of regularized high-dimensional methods. Extensive data analysis shows that integrative analysis can statistically and biologically outperform the existing single-dataset/single-dimensional anlaysis.