[Song Young-doo, Edaily Reporter] Syntekabio announced on the 11th that its core platform DeepMatcher is setting the direction for a new paradigm in AI-driven drug discovery, cognitive AI drug development.

Drug discovery is essentially the process of identifying the best-fitting candidate molecule for a specific protein among countless possibilities. However, manually reviewing all potential candidates requires many years and costs ranging from hundreds of millions to several billions of dollars. For this reason, many companies seek to adopt AI technology, but in practice, few platforms deliver meaningful results.

So far, most AI systems have remained little more than “ultrafast calculators.” Even docking billions of compounds against proteins is inefficient when hardware costs and electricity are considered. In other words, such approaches automate calculations rather than address the fundamental challenges of drug discovery.

According to Syntekabio, the essence of AI drug discovery lies in “cognition and interpretation.” A true AI platform should be able to understand the structural and biological context of proteins and rapidly narrow down rational candidate pools based on that understanding.

For example, if AI recognizes a particular protein as an RNA-binding protein, it should infer the possibility of sub-pockets capable of interacting with RNA bases, and among billions of compounds, prioritize recommending derivatives containing RNA base analogues. This represents AI substituting for human researchers’ cognitive capabilities and constitutes a commercially meaningful AI drug discovery platform.

DeepMatcher embodies this cognitive approach. The platform integrates △RVR-FP(Record Virtual Rack Fingerprint) an indexing tool that analyzes binding sites based on 3D protein structures and extracts key drug scaffolds. △LM-VS(Language Model Virtual Screening) an ultrafast search engine capable of screening over 10 billion compounds. △3D-CNN(3D Convolutional Neural Network) and MD(Molecular Dynamics) simulations tools to refine the interactions of derived candidate molecules.

A Syntekabio official stated, “With these capabilities, we can go beyond simple docking calculations to cover the entire process from candidate search and derivative design to precision validation. Syntekabio has already proven this value through multiple successful joint projects with global pharmaceutical companies.”

When compared globally, the differentiation is also clear. Google DeepMind’s AlphaFold made a breakthrough in predicting protein structures, but it has not extended to the candidate search and validation stages required for drug development.

In contrast, Syntekabio possesses a comprehensive platform combining RVR-FP, LM-VS, 3D-CNN, and MD simulations, positioning itself as an industry leader. While the platform is currently semi-automatic, the company aims to evolve it into a fully automated system, ultimately establishing an AI drug discovery platform on par with OpenAI.

Syntekabio concluded, “We will continue to move beyond computational AI toward AI platforms that understand and interpret biological properties, dramatically accelerating drug discovery and increasing success rates.”