[Edaily Reporter KIM JI-WAN ] As an era dawns in which artificial intelligence (AI) designs hundreds of thousands of new drug candidates, wet labs—which validate these candidates through actual experiments—are emerging as a new bottleneck in AI-driven drug development. Analysis suggests that #PROTEINA CO., LTD. is targeting this market specifically based on its protein-protein interaction (PPI) analysis technology.

In a report released on the 2nd, IV Research noted, “The key bottleneck in future AI-driven drug development will be the limited capacity for experimental validation, rather than the AI models themselves,” drawing attention to the potential for PROTEINA CO., LTD. to expand its AI Biofoundry business. The firm did not provide an investment opinion or a target stock price.

Even if AI Designs Hundreds of Thousands of Candidates, They Must Ultimately Undergo ‘Experiments’

AI-driven drug discovery technology has advanced rapidly in recent years. With the emergence of open-source models—such as AlphaFold3, Boltz-1, Protenix-v1, and OpenDDE—that predict protein structures and antigen-antibody binding, the barriers to entry for generating candidate sequences have significantly lowered.

The issue is whether AI-generated candidate compounds can actually become commercial drugs. Antibody candidates must not only bind well to their targets but also be verified for mass-producibility, resistance to aggregation, and stability at specific temperatures. This is known as “developability.”

IV Research analyzed that while technologies for predicting binding affinity and protein structures are improving rapidly, data on developability—such as productivity, aggregation potential, and stability—remains relatively scarce, leading to low accuracy in AI predictions. Ultimately, this means that the more candidate compounds AI generates, the greater the scarcity of infrastructure needed to validate them through actual experiments.
PROTEINA CO., LTD.’s AI Biofoundry targets precisely this issue. It is a service that expresses AI-designed antibody candidates on a large scale in actual wet labs, evaluates characteristics such as binding affinity and productivity, and selects final candidates.

The company plans to offer the processes—candidate generation, expression, evaluation of binding affinity and productivity, and candidate screening—which it has utilized for its internal drug discovery through its existing PPI Landscape business, as a service to external customers. The company aims to expand its current monthly analytical capacity (CAPA) of approximately 40,000 tests to 1 million tests per month by 2028—a roughly 25-fold increase—through full wet lab automation.

The company is also shifting its business structure away from a model that relies solely on AI models to select candidates toward a “Lab-in-the-Loop” approach—a structure that involves iterating between AI-driven design and validation through actual experiments—to enhance the efficiency of new drug development.

Collaboration with Twist… A Test Case for Commercializing the Biofoundry Business

The collaboration with Twist Bioscience is cited as

the

first

test case

for

commercialization

. IV Research forecasts that once both companies complete platform validation, it is likely to lead to a formal agreement. If such an agreement is signed, the expected structure would be for Twist to produce candidate DNA sequences, while PROTEINA CO., LTD. handles everything from protein expression to characterization.

Existing businesses are also linked to the AI biofoundry. PROTEINA CO., LTD.’s core businesses are PPI Landscape and PPI PathFinder. PPI Landscape is a business that uses the Single-Molecule Protein Interaction Analysis (SPID) platform to measure and optimize the characteristics of antibody candidates on a large scale. The company has signed a joint development and technology transfer option agreement with Samsung Bioepis and is currently conducting an evaluation.

PPI PathFinder identifies and validates biomarkers by analyzing changes in protein-protein interactions in clinical patient samples. The company is currently providing analysis and validation services for BCL-2 biomarkers in blood cancers and is expanding its scope to include inflammatory and neurological diseases. In terms of its own new drug candidates, it is developing “PRT-101,” a treatment for osteoarthritis, and “PRT-1309,” a treatment for obesity and metabolic disorders.

PROTEINA CO., LTD.’s investment appeal ultimately lies not in the “AI model competition” itself, but in its validation infrastructure—which filters the vast number of candidate compounds generated by AI into actual drugs. This is because as the number of AI-driven drug discovery companies increases and the speed of candidate compound generation accelerates, the demand for wet-lab validation is also expected to grow.

IV Research stated, “The key bottleneck in AI-driven drug discovery will be the ability to conduct actual experimental validation, rather than the models themselves.” As AI accelerates the generation of drug candidates, PROTEINA CO., LTD. plans to expand its business model to become a “validation factory” for those candidates.