Target-based drug discovery has become a central strategy in modern pharmaceutical research, but the selection of effective molecular targets remains a major bottleneck. Transcriptomic and proteomic technologies can generate hundreds or thousands of disease-associated genes and proteins from clinical samples, yet only a small fraction may be suitable for therapeutic development. Existing computational prioritization methods often favor well-studied proteins with known disease annotations, which may limit their ability to identify original targets. The research team previously developed OTTM, a tool for drug repositioning that focuses on targets with approved or clinical drugs. However, many omics-derived proteins lack active compounds, creating the need for a complementary approach that can prioritize unexplored but potentially druggable targets.
The study published in Targetome on 30 April 2026 by Hao Zhang's team, Shanghai University of Traditional Chinese Medicine, reports that OPTAR can identify novel, druggable disease targets from omics data and validates UBE2J1, KDELR3, and VTI1A as potential HCC-related targets.
To build OPTAR, the researchers designed a multi-step computational workflow. First, users input a list of differentially expressed genes or proteins and a disease keyword. OPTAR screens candidate proteins against PubMed abstracts to exclude proteins already reported to be associated with the disease of interest. It also uses drug-target information from the Therapeutic Target Database to retain proteins without approved or clinical drugs, focusing on candidates with high originality. Next, OPTAR infers disease relevance indirectly through protein-protein interaction (PPI) networks derived from STRING. For each candidate protein, the tool calculates a disease-correlation score based on how often its interacting partners appear in disease-related PubMed abstracts. This allows the tool to prioritize proteins that are not directly reported in the disease literature but are biologically connected to disease-associated networks. Finally, OPTAR evaluates potential druggability by analyzing AlphaFold-predicted protein structures to identify binding pockets suitable for small molecules. The team tested OPTAR using HCC omics data previously used in OTTM research. With “hepatocellular carcinoma” and “cell cycle” as search keywords, OPTAR generated two lists of proteins that lacked known drugs and prior disease associations: 383 proteins for HCC and 542 proteins for cell cycle. The intersection of the top 200 candidates from both lists yielded 42 proteins. After PPI-based ranking and pocket assessment, the researchers selected UBE2J1, VTI1A, and KDELR3 for experimental validation. In HepG2 and Huh7 liver cancer cells, siRNA-mediated knockdown of all three genes reduced cell viability and suppressed migration and invasion. UBE2J1 was further examined because recombinant protein was commercially available. Public database analysis showed that UBE2J1 expression was elevated in liver hepatocellular carcinoma tissues compared with normal liver tissues. UBE2J1 knockdown also altered apoptosis-related markers and disrupted cell-cycle distribution. The researchers then performed virtual screening of 4,654 traditional Chinese medicine compounds against UBE2J1, selected 30 compounds for surface plasmon resonance testing, and found that 22 directly bound UBE2J1. Echinacoside and Butein showed the strongest binding affinities, while Butein markedly inhibited HepG2 cell viability.
Overall, the study presents OPTAR as a computational strategy for discovering novel therapeutic targets from omics data, with a focus on proteins that are both underexplored and potentially druggable. By validating candidate targets in liver cancer cells and identifying small molecules that bind UBE2J1, the work demonstrates a route from omics-based prediction to experimental target confirmation and compound discovery.
Yuan X, Zhou S, Yu J, Wang M, Luo C, et al.
OPTAR: a computational tool for target discovery based on disease correlation inference from literature of interacting proteins.
Targetome 2(2): e018 doi: 10.48130/targetome-0026-0017