3Billion will present research on AI-based genome interpretation and drug target discovery for rare disease diagnosis and new drug development at the International Machine Learning (ML) Conference.
3Billion announced on Tuesday that four papers have been accepted for a workshop at ICML 2026, an international conference on machine learning to be held at COEX in Seoul from July 6th to 11th.
ICML is considered one of the world’s three major AI and ML conferences, along with NeurIPS and ICLR, and serves as a forum for AI researchers and experts from around the world to share the latest research results and technology trends in machine learning. At the ICML 2026 workshop, 3Billion will present a total of four papers introducing gene mutation interpretation techniques for rare disease diagnosis and drug target discovery research using patient genomic data.
Three of the papers were accepted for the “GenBio” workshop, which deals with the application of generative and agentic AI to biological research. GenBio is a workshop where researchers from global AI companies and leading universities and research institutions discuss AI technologies in the life sciences field. The contents of the accepted papers are as follows. ▲ Discovery of suppressor variants (AnomalyModifier) using data from familial hypercholesterolemia patients. ▲Research on loss-of-function mutation interpretation using protein language models (ESM-2).
“AIVARI Agent” developed by 3Billion is an AI agent that identifies genetic mutations that require clinical reporting. This AI agent comprehensively analyzes a large body of evidence, including variant pathogenicity, inheritance patterns, and literature, and provides interpretive hypotheses for candidate variants to assess their disease association. AI supports the process of interpreting genetic mutations, which was previously performed by experts reviewing multiple pieces of evidence, improving diagnostic accuracy, reliability, and consistency.
“AnomalyModifier” is an AI model that discovered suppressor variants from data on familial hypercholesterolemia patients. AIVARI Agent improves diagnostic accuracy and reliability, while AnomalyModifier extends patient genomic data accumulated through diagnosis to drug target discovery. Suppressor mutants are mutants that reduce the expression of symptoms or the effects of a disease even in the presence of a disease-causing mutant, and genes that such mutants act on can be candidates for drug discovery targets. The research team used anomaly detection techniques to address the problem of drug target discovery for rare diseases where ground truth data is lacking, and demonstrated that patient genomic data can serve as a resource for finding potential therapeutic targets.
Research using protein language models (ESM-2) to interpret loss-of-function variants was also accepted at the GenBio workshop. The research team proposed a framework that decomposes the complex internal representations of protein language models into biologically meaningful conceptual units in order to predict loss-of-function variants in an explainable manner.
Additionally, one paper was accepted for the “FM4LS” workshop. The FM4LS workshop tackles multimodal foundational models and large-scale language models in the life sciences, and includes researchers from universities and research institutes around the world, as well as leading AI companies such as Google DeepMind and Microsoft Research. In an approved study, 3Billion proposed a multimodal AI framework that combines protein and DNA language models to predict the pathogenicity of genetic variants. As a result, the context of mutation interpretation, which is difficult to grasp with protein sequence information alone, is supplemented with DNA information, improving the accuracy and reliability of clinical mutation interpretation.
3Billion CEO Kim Chang-won said, “The acceptance of this paper at this ICML workshop is an achievement in which 3Billion’s AI genome interpretation technology has been recognized on the world stage.” “We will continue to advance our AI technology, which encompasses both genetic diagnostics and patient data-based target discovery, to help expand diagnostic and treatment opportunities for patients with rare diseases.”
