BackgroundMajor depressive disorder and bipolar disorder are prevalent and debilitating psychiatric disorders that are difficult to distinguish, as their diagnosis is based on behavioural observations and subjective symptoms. Quantitative protein profile analysis might help to objectively distinguish between these disorders and increase our understanding of their pathophysiology. Thus, this study was conducted to compare the peripheral protein profiles between the two disorders.MethodsSerum samples were collected from 18 subjects with major depressive disorder and 15 subjects with bipolar disorder. After depleting abundant proteins, liquid chromatography-tandem mass spectrometry (LC-MS/MS) and label-free quantification were performed. Data-dependent acquisition data were statistically analysed from the samples of 15 subjects with major depressive disorder and 10 subjects with bipolar disorder who were psychotropic drug-free. Two-sided t-tests were performed for pairwise comparisons of proteomes to detect differentially-expressed proteins (DEPs). Ingenuity Pathway Analysis of canonical pathways, disease and functions, and protein networks based on these DEPs was further conducted.ResultsFourteen DEPs were significant between subjects with major depressive disorder and those with bipolar disorder. Ras-related protein Rab-7a (t = 5.975, p = 4.3 × 10− 6) and Rho-associated protein kinase 2 (t = 4.782, p = 8.0 × 10− 5) were significantly overexpressed in subjects with major depressive disorder and Exportin-7 (t = -4.520, p = 1.5 × 10− 4) was significantly overexpressed in subjects with bipolar disorder after considering multiple comparisons. Bioinformatics analysis showed that cellular functions and inflammation/immune pathways were significantly different.ConclusionsRas-related protein Rab-7a, Rho-associated protein kinase 2, and Exportin-7 were identified as potential peripheral protein candidates to distinguish major depressive disorder and bipolar disorder. Further large sample studies with longitudinal designs and validation processes are warranted.