Newsletter

POSTECH develops AI that accurately predicts oncogenic mutations… Increasing the possibility of early cancer diagnosis

POSTECH (President Moo-Hwan Kim) announced on the 23rd that a research team led by Professor Sang-Wook Kim from the Department of Life Sciences, Dr. Dong-Hyo Kim, and Dr. Do-Yeon Ha has developed a machine learning model that can accurately predict whether mutations in a patient’s gene can cause tissue-specific cancer.

Identifying cancer-causing mutations (driver mutations) is important for elucidating different pathological mechanisms across a variety of tumors and providing patient-specific treatment opportunities. The research team devised new capabilities based on sequence co-evolution analysis to identify driver mutations by cancer type and build machine learning (ML) models with state-of-the-art performance. It is a method that can detect cancer driver mutations through accumulated data from 28,000 tumor samples across 66 types of cancer.

The research team developed a model that predicts the oncogenic potential of mutations through protein sequence analysis and machine learning, and was able to achieve high accuracy and sensitivity compared to existing models. In addition, artificial intelligence (AI) was trained using protein sequence evolution analysis, which has not been used in previous studies, to find protein residues (specific units in the polymer chain of polysaccharides, proteins, and nucleic acids) or mutations that can cause certain carcinomas. .

Professor Sang-wook Kim's team at POSTECH developed an AI that accurately predicts cancer-causing mutations.  It is expected to be useful in the early diagnosis of cancer.  Images related to research.

The mutations discovered through the study have been shown to disrupt tissue-specific protein interaction networks, leading to certain types of carcinoma. The results of this research are expected to help find effective cancer prevention and treatment methods by combining early cancer diagnosis technology and discovering new cancer treatment targets.

Professor Kim Sang-wook said, “Using this technology, it is possible to find new oncogene mutations that could not be detected before, and it will help establish cancer diagnosis and treatment strategies that different from existing technologies.”

The results of this research, conducted with the support of the POSTECH Medical Device Innovation Center, the Graduate School of Artificial Intelligence, and the Korea National Research Institute, were recently published in ‘Briefings in Bioinformatics’, an authoritative journal in the field of bioinformatics.

Pohang = Reporter Jeong Jae-hoon jhoon@etnews.com