Precision Analysis of Protein Modifications Using AI Prediction Technology [Reading Science]

A Step Closer to Uncovering the Causes of Intractable Diseases

By precisely identifying extremely rare protein modifications inside cells using artificial intelligence (AI) prediction technology, researchers have taken a step closer to uncovering the molecular-level causes of intractable diseases such as cancer.

On January 29, a research team led by Dr. Lee Cheolju at the Chemical Biology Research Center of the Korea Institute of Science and Technology (KIST) announced that they had developed a technology that accurately detects rare protein modifications, which had been difficult to distinguish with conventional analysis methods, by utilizing an AI learning model.

Overview of AI-Based Formula Protein Discovery Technology Development Research. Graphic provided by the research team

Intractable diseases such as cancer are closely linked to subtle protein changes that occur as cells experience stress. However, these modifications occur at a very low frequency and their signals are similar to false positives, making it difficult for existing mass spectrometry techniques to accurately identify them. As a result, there has been a growing demand for new analytical technologies capable of tracing the fundamental causes of diseases at the molecular level.

The modification targeted by the research team is "arginylation," which acts as a signal for regulating protein function or degradation. Abnormalities in this process can lead to neuronal cell damage or cancer development. However, the extremely low abundance of arginylation in living organisms has made it difficult to distinguish true signals from false ones. To address this challenge, the team adopted a novel approach by first training the AI with false signals that closely resemble true ones.

As a result, they succeeded in eliminating approximately 90% of the false signals detected in previous analyses and identified a total of 134 actual arginylation modification sites. In particular, by applying transfer learning techniques, they demonstrated that even with a small amount of data, it is possible to precisely analyze rare protein modifications. Analysis of cells under stress conditions confirmed arginylation modifications in certain proteins related to cellular energy production, providing new clues about cancer cell metabolism.

This technology integrates the entire process from protein modification discovery to primary verification into a single AI-based analytical system, which is expected to significantly reduce research costs and time in drug development and bio-research fields. When applied to patient blood or tissue analysis, it also has great potential as a foundational technology for early diagnosis and precision medicine research by enabling faster and more accurate detection of disease-related protein changes.

Dr. Lee Cheolju of KIST stated, "This achievement boldly introduced AI into an area that has remained a limitation in previous research. With this domestically developed, world-class AI-based proteome analysis technology, we aim to contribute to the expansion of AI-driven proteomics research."

This research was conducted with support from the Ministry of Science and ICT, through KIST's major projects, individual basic research projects, and the Bio Research Data Utilization Infrastructure Project. The research findings were published in the international journal Nature Communications.