AbstractMajor depressive disorder (MDD) is a common but serious psychiatric disorder with significant levels of morbidity and mortality. Recent genome-wide association studies (GWAS) on common variants increase our understanding of MDD; however, the underlying genetic basis remains largely unknown. Many studies have been proposed to explore the genetics of complex diseases from a viewpoint of the “missing heritability” by considering low-frequency and rare variants, copy-number variations, and other types of genetic variants. Here we developed a novel computational and statistical strategy to investigate the “missing heritability” of MDD. We applied Hamming distance on common, low-frequency, and rare single-nucleotide polymorphism (SNP) sets to measure genetic distance between two individuals, and then built the multi-dimensional scaling (MDS) pictures. Whole-exome genotyping data from a Los Angeles Mexican-American cohort (203 MDD and 196 controls) and a European-ancestry cohort (473 MDD and 497 controls) were examined using our proposed methodology. MDS plots showed very significant separations between MDD cases and healthy controls for low-frequency SNP set (P value < 2.2e−16) and rare SNP set (P value = 7.681e−12). Our results suggested that low-frequency and rare variants may play more significant roles in the genetics of MDD.