AsianScientist (Oct. 03, 2025) – Breakthrough systems like DeepMind’s AlphaFold have demonstrated the remarkable ability of artificial intelligence (AI) in predicting protein structures, marking a pivotal milestone in computational biology. Now, researchers are pursuing the next frontier in which AI is used to understand protein interactions in order to predict function from sequence.
The researchers generating AI models, however, are often computational experts rather than wet lab specialists. This creates a knowledge gap that can hinder the transition from digital prediction to experimental validation. As AI models generate increasingly sophisticated models for protein designs, there is an urgent need for greater integration and skillset sharing between computational and biological experts.
Leading biotechnology company Twist Bioscience is making significant strides to address this bottleneck. By responsibly providing high-quality, customizable DNA synthesis in the formats and scales AI developers need, Twist Bioscience translates customers’ AI-powered designs into physical sequences for experimental validation.
“One major challenge is the vastness of the protein sequence space; finding a high-functioning enzyme or antibody can be like finding a needle in a haystack. AI helps by intelligently narrowing down candidates, but the real breakthrough comes when you can quickly and accurately synthesize all those AI-predicted designs for testing in the exact DNA format you need,” explained Dr. Julian Jude, Director of Emerging Applications SynBio at Twist Bioscience.
Antibody discovery, for instance, often requires researchers to screen tens of millions of variants in order to identify improved candidates. Using AI helps reduce this number of variants by several orders of magnitude to thousands of defined sequences that are most likely to exhibit desired functions, which could accelerate the discovery pipeline.
However, the predictive power of AI is only as valuable as the researchers’ ability to physically test those predictions. As such, Twist Bioscience leverages its DNA synthesis platform to manufacture products and provide experimentation services that support high-throughput protein discovery workflows.
Among them are Twist Multiplexed Gene Fragments (MGFs), which enable researchers to screen thousands of gene fragments simultaneously, each up to 500 base pairs in length. Unlike traditional synthesis methods that produce one fragment per tube, MGFs deliver entire libraries of variants in a single pooled format, making it ideal for testing AI-designed protein libraries quickly and at scale.
“At 500 base pairs, MGFs are much longer than standard oligonucleotides and can cover whole protein domains or even full proteins,” noted Dr. Jude. “Importantly, they aren’t constrained by high guanine-cytosine (GC) content or repetitive sequences that are problematic for other synthesis methods. This is particularly crucial for AI-driven campaigns where sequences need to be represented exactly as designed by the algorithm.”
Complementing the MGF technology, Twist Oligo Pools provide highly diverse collections of single-stranded DNA oligonucleotides ranging from 20 to 300 nucleotides. These pools can contain hundreds of thousands of unique sequences, enabling researchers to encode peptide libraries, variable antibody regions or regulatory elements for comprehensive screening studies. They serve as a cost-effective, one-stop solution to generate huge libraries of precisely defined sequences.
By delivering high-throughput, high-quality DNA in the formats researchers need, Twist Bioscience supports AI-powered projects across multiple research areas, helping teams move from in silico designs to experimental validation.
At the University of Washington’s Baker Lab, for instance, researchers used AI-guided protein design software combined with Twist Bioscience’s MGF technology to discover novel antibody fragments. Through this approach, they were able to design and validate new antibody binders. In separate experiments, researchers at the Baker Lab also used Twist Oligo Pools to design new proteins that neutralize the virulence factors causing botulism and influenza.
In agricultural biotechnology, Phytoform Labs leveraged Twist Oligo Pools to screen thousands of AI-designed DNA sequences, streamlining trait engineering for crop improvement. The approach allowed them to optimize research resources while accelerating the development of more resilient food crops.
Meanwhile, researchers at Yonsei University in South Korea combined massive oligo libraries with deep learning models to predict gene editing efficiency with unprecedented accuracy, demonstrating how AI and synthetic biology can synergize to advance therapeutic applications.
The combination of AI prediction and rapid DNA synthesis is delivering tangible benefits in terms of the speed and success rate of biological engineering challenges too. Traditional antibody discovery, for example, typically requires 12 to 18 months—AI-assisted approaches can potentially reduce this timeline to three to six months, according to Dr. Jude. Companies using AI-driven methods report success rates that are three to four times higher than conventional approaches.
“By marrying AI’s predictive power with rapid DNA synthesis, projects that might have taken months or years can reach milestones in weeks,” Dr. Jude observed. “Customers can explore much broader design spaces and solve tougher problems, such as finding binders to ‘undruggable’ targets or enzymes for novel chemical reactions, because AI can propose creative solutions and our platforms can build them.”
The demand for high-throughput, high-accuracy DNA synthesis will only grow. Twist Bioscience is taking a proactive approach, continuously expanding its capabilities to meet the evolving needs of AI-driven research. The company’s vision aligns with the broader trend toward automated discovery systems where AI and synthetic biology work in concert.
As Dr. Jude reflected, “By maintaining a close dialogue with the research community and investing in forward-looking research, Twist Bioscience aims to not react to trends but help set them. In an era where AI is transforming biopharma, we want to be the bridge that enables researchers to fully realize the potential of AI in developing better medicines and biotechnology solutions.”
Twist Biopharma Solutions, a division of Twist Bioscience, exemplifies how AI has transformed biologics discovery, as the company integrated AI machine learning technology across its antibody discovery and optimization platforms. The company enhances the success rate of an antibody discovery campaign by leveraging AI to generate antibody repertoires that are enriched for fully human sequences specific to pre-defined target epitopes. With AI, it has also shortened the antibody humanization and affinity maturation timeline by three folds, from five months to six weeks.
Complementing these AI-driven approaches, Twist Bioscience provides high-throughput antibody production and characterization capabilities, including comprehensive developability and immunogenicity profiling. These ensure that lead candidates are not only potent, but also well-suited for downstream therapeutic development.
For more information on Twist Bioscience’s DNA synthesis platform, contact them here.
—
Source: Twist Bioscience, Shutterstock
Disclaimer: This article does not necessarily reflect the views of AsianScientist or its staff.