Major depressive disorder (MDD) is a debilitating psychiatric disorder characterised by a complex underlying biology and poor response to pharmacological antidepressant strategies. Given the heterogeneity of MDD and the diverse range of available treatment options, there is an increasing desire to develop and implement precision medicine approaches to tailor existing treatment strategies to the biological system of the individual. In this thesis, high-resolution omics data (connectomics [fMRI], metabolomics [1H NMR] and immunomics [inflammatory cytokines]) collected from the Canadian Biomarker Integration Network in Depression (CAN-BIND) study has been integrated to facilitate the deep phenotyping of MDD. In addition, this approach has been used to predict the treatment response to two common antidepressant drugs, monotherapy with the selective serotonin reuptake inhibitor (SSRI) escitalopram (10-20 mg) or combination therapy with escitalopram and the dopaminergic antipsychotic aripiprazole (2-10 mg). This approach identified a multi-modal panel of sex-specific biomarkers of MDD and treatment response, highlighting a strong immunometabolic component in depressed males, but not females. Unsupervised clustering methods indicated the superiority of biological (neuroimaging) over symptom-based (clinical questionnaires) data for the stratification of patients into MDD subtypes with differential response to treatment. More importantly, a set of multi-modal, sex-specific biomarkers were identified that predicted treatment response with escitalopram monotherapy (84.7% accuracy) or aripiprazole augmentation (88.5% accuracy). In addition to highlighting potential new aspects of the biology of MDD (e.g. relevance of lipoprotein size and density for their relation to depression), this work is one of the first attempts to apply systems biology approaches to high-resolution biological data from a large clinical trial to predict later treatment outcome. With the validation of the findings presented in this thesis in independent cohorts, and with further development of omics technologies, leading to cheaper and high-throughput screening of the patient population, pre-dose biomarkers have the potential to achieve personalised treatment. Each year, escitalopram and aripiprazole are prescribed to an estimated 26 million and 7 million individuals respectively, and over one third of them do not respond. Thus, being able to predict response to antidepressant medication from baseline biomarkers has enormous clinical and socioeconomic benefits.