The premise behind the work rests on a growing body of evidence that depression is not purely a disorder of neurotransmitters. Brain imaging studies have repeatedly shown altered energy metabolism in depressed patients, and post-mortem tissue analyses have hinted at mitochondrial dysfunction in neural circuits that govern mood. Mitochondria do far more than generate ATP; they regulate calcium signaling, produce reactive oxygen species, and even help decide when a cell should die. When these processes falter, neurons may become less resilient to stress, and inflammatory signaling can spiral. What has been missing is a precise, reproducible set of genes that ties mitochondrial metabolism to the clinical reality of major depressive disorder, and that is precisely the gap the new study set out to fill.

The team, led by Lianyong Zou and corresponding author Jie Li of Tianjin Anding Hospital and the Shandong Mental Health Center, began with a classic bioinformatics strategy. They scoured the Gene Expression Omnibus, a public repository maintained by the National Center for Biotechnology Information, for datasets comparing gene expression in blood or brain samples from people with major depressive disorder against healthy controls. From these datasets they extracted differentially expressed genes, the molecular transcripts that rise or fall significantly in depression. In parallel, they compiled a reference list of 1,234 genes known to participate in mitochondrial metabolism, drawn from public pathway databases. The intersection of these two lists, genes that are both dysregulated in depression and directly involved in mitochondrial metabolism, formed the starting pool of candidates.

That intersection alone would have produced a long list, so the researchers turned to machine learning to whittle it down. Using expression-screening algorithms trained on the depression datasets, they iteratively ranked and filtered the candidate genes, searching for those whose expression patterns best discriminated depressed patients from controls. The process converged on three genes: ALDH3B1, ALOX15B, and UPB1. Each has a distinct biochemical role. ALDH3B1 encodes an aldehyde dehydrogenase involved in detoxifying reactive aldehydes, byproducts of lipid peroxidation that can damage proteins and DNA. ALOX15B encodes a lipoxygenase enzyme that oxidizes polyunsaturated fatty acids into signaling lipids, many of which are potent mediators of inflammation. UPB1, or beta-ureidopropionase, participates in the breakdown of pyrimidine bases, connecting the gene to nucleotide metabolism and, indirectly, to cellular energy economics.

To test whether these three genes could serve as a practical diagnostic signature, the team built a nomogram, a statistical tool that combines the expression values of multiple genes into a single risk score. In the training dataset, the three-gene nomogram showed moderate discriminatory performance, meaning it could separate depression cases from controls better than chance, though not with the precision required of a standalone clinical test. The authors are careful on this point: the nomogram is a proof of concept, a demonstration that a compact mitochondrial gene panel carries real signal, rather than a finished diagnostic instrument. Still, in a field where biomarkers for depression have been notoriously hard to pin down, even moderate performance from a three-gene panel is noteworthy.

The researchers then asked what biological programs these genes travel with. Using gene set enrichment analysis, a technique that examines whether entire pathways shift in concert with a gene of interest, they found that ALDH3B1, ALOX15B, and UPB1 correlated with transcriptional programs involving lysosomal function, protein turnover, ribosomal processes, cellular signaling, and energy metabolism. This is a coherent picture: lysosomes are the cell’s recycling centers, ribosomes its protein factories, and mitochondria its power grid, and all three must be coordinated for a cell to stay healthy. A simultaneous perturbation across these systems is consistent with the idea that depression involves a broad breakdown in cellular housekeeping, not a single faulty switch.

One of the more striking findings came from immune infiltration analysis. By estimating the relative abundance of different immune cell types from gene expression data, the team detected significant differences in six immune-cell signatures between depressed patients and controls. Several myeloid-cell signatures, which encompass monocytes, macrophages, and related innate immune cells, correlated with the expression of the candidate genes. This dovetails with a decade of research linking depression to low-grade systemic inflammation, and it suggests that the mitochondrial genes identified here may sit at the junction between metabolic dysfunction and immune activation. If lipid-peroxidation enzymes like ALOX15B are upregulated in depression, they could generate oxylipin signaling molecules that both recruit myeloid cells and alter neuronal function, offering a mechanistic bridge between the two processes.

The study also ventured into therapeutics. An exploratory drug-gene interaction analysis identified approved medications predicted to interact with each candidate gene: four drugs for ALDH3B1, eighteen for ALOX15B, and two for UPB1. The authors frame this as hypothesis generation rather than treatment guidance, and appropriately so. Drug-gene interaction databases catalog known pharmacological targets, but whether any of these compounds could meaningfully modulate the course of depression through these genes remains entirely untested. Still, the exercise illustrates a modern trend in psychiatry research: using computational pipelines to surface repurposing candidates that can then be prioritized for laboratory and, eventually, clinical scrutiny.

The critical step was experimental validation. The team performed reverse transcription quantitative PCR, a sensitive technique for measuring gene expression, on samples from depressed patients and healthy controls, with ethical approval from the Shandong Mental Health Center and written informed consent from all participants. The results were encouraging but nuanced. ALDH3B1 was significantly elevated in depression, with a p-value of 0.0138. ALOX15B showed an even stronger signal, with a p-value below 0.0001, making it the standout candidate of the study. UPB1 trended upward with a fold change of 1.3245 but did not reach statistical significance, with a p-value of 0.4842. In their conclusions, the authors classify ALDH3B1 and ALOX15B as preliminary candidate genes supported by both bioinformatics and experimental observation, while UPB1 remains a bioinformatics-derived candidate awaiting independent validation.

The researchers are candid about the limitations. The RT-qPCR cohort was small, and the authors explicitly state that the candidates require further investigation in larger independent cohorts. Blood-based gene expression, the most accessible sample type for psychiatric studies, may not perfectly mirror what happens in the brain, and depression is clinically heterogeneous, encompassing patients whose biology may differ substantially. Yet the study’s architecture, from public data mining through machine learning to molecular validation, offers a template for how psychiatric biomarker research can proceed in the era of big data. If larger studies confirm that ALOX15B and ALDH3B1 are reliably dysregulated in depression, they could become part of a much-needed molecular toolkit: objective measures to complement subjective diagnostic interviews, and potential targets for therapies aimed at the metabolic and inflammatory undercurrents of the illness. For now, the message is one of cautious excitement. The mitochondria, long dismissed as mere cellular plumbing, are increasingly looking like central characters in the biology of mood, and three genes may be the first reliable thread in a story that is only beginning to unravel.

Subject of Research: Mitochondrial metabolism-related candidate genes associated with major depressive disorder identified through bioinformatics and RT-qPCR validation

Article Title: Mitochondrial metabolism-related candidate genes in major depressive disorder: integrative bioinformatics analysis and RT-qPCR assessment

Article References: Zou, L., Zhao, Y., Han, C., Liu, J., Liu, H., Chen, Z., Li, Q., & Li, J. (2026). Mitochondrial metabolism-related candidate genes in major depressive disorder: integrative bioinformatics analysis and RT-qPCR assessment. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08636-3

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08636-3

Keywords: major depressive disorder, mitochondrial metabolism, ALDH3B1, ALOX15B, UPB1, bioinformatics, machine learning, RT-qPCR, gene expression, immune infiltration, biomarkers, BMC Psychiatry