[Edaily Reporter Hong Ju-yeon ] Hanmi Pharmaceutical ( HanmiPharm(128940)) is utilizing its proprietary artificial intelligence (AI) platform to identify novel, innovative drugs for obesity and next-generation muscle-building therapeutic candidates, and is currently conducting clinical development.

HanmiPharm announced on the 24th that it presented research results on a muscle-strengthening-based obesity drug pipeline developed using its proprietary AI platform “HARP-pSAR” at the International Society for Molecular Biology and Intelligent Systems (ISMB) conference, held in Washington, D.C., from the 12th to the 16th.

According to HanmiPharm, this presentation drew attention because the AI platform went beyond being a simple predictive tool to actually lead to the identification of clinical-stage candidate compounds. Using HARP-pSAR, HanmiPharm identified “LA-UCN2 (code name HM17321),” the world’s first innovative obesity drug that simultaneously achieves weight loss and muscle gain; it is currently undergoing a Phase 1 clinical trial in the U.S. In June of this year, at the American Diabetes Association (ADA) conference, the company also unveiled “LA-MSTN (code name HM500197),” the world’s first next-generation muscle-building therapy based on a peptide-based myostatin inhibition mechanism.

HM17321 was designed as a UCN2 analog that selectively targets the CRF2 receptor rather than incretin receptors such as GLP-1. HanmiPharm explains that selectively targeting the CRF2 receptor can directly lead to fat reduction, muscle growth, and improved muscle function. The key development challenge was to meticulously optimize the sequence so that the compound binds strongly to the target receptor, CRFR2, while showing no activity toward the off-target receptor, CRFR1.

HARP-pSAR (Hanmi AI-driven Research Platform – Protein Sequence-Activity Relationship) was designed to overcome the lack of experimental data, a major obstacle in the early stages of new drug development. By applying quantization technology, it reduces computational costs and can run even on local workstation environments. Using only the results of dozens of internal experiments, it predicts activity against target and off-target receptors with high accuracy. It can even predict the activity of new sequences not included in the training data and suggest optimal amino acid residue positions, thereby significantly reducing trial and error and costs in the candidate compound development process.

Choi In-young, Head of the Future Growth Division (Vice President) at HanmiPharm, stated, “This is a pioneering example demonstrating that AI can serve as a practical research partner—going beyond a simple data analysis tool to suggest design directions for new drug candidates and reduce trial and error in the development process.” He added, “It is particularly significant that we presented these results at ISMB, as the AI-based design outcomes have led to the actual identification of drug candidates and clinical development.”