Changing How We Explore Sequence Space In The AI Era

With the recent launch of our breakthrough Gene Pools product, an old question at Twist has gained new life: “What would you do with unlimited DNA?”

In the summer of 2015, I stood on stage at SynBioBeta London and asked the audience this one simple question. It was there that we officially launched Twist Bioscience alongside our first ever product: Cloned Genes. The pitch was simple: We have miniaturized oligo synthesis with our new proprietary silicon DNA synthesis platform. We can make ten-thousandfold more oligos per run than anyone else. We make them better than anyone else. Now scientists should go and use this now-abundant raw material to improve the world.

When we launched, we were able to offer gene synthesis for just 9¢ per base, meaning the price of ordering DNA more-than-halved overnight. Scientists could suddenly start ordering and testing more constructs. For the Synthetic Biology field, the promise of scale meant that a vast sequence space was, for the first time, affordable and accessible.

Since then, innovation has always been core to our ethos at Twist, and that means returning to our base offering and asking “how can we make this even better?”

This requires more than creativity, though. It demands a deep understanding of the ways in which molecular biology is advancing and the barriers that slow that advance. AI, for example, has significant potential in helping researchers design and optimize new proteins. But, in these efforts, the generation of training datasets and in vitro model validation have quickly emerged as choke-points that limit the development of robust protein prediction models. Specifically, researchers are struggling to get the sheer volume of bespoke synthetic DNA sequences needed to power the screens used to train and validate models, particularly at a reasonable price point.

Put simply, most DNA providers are ill equipped to make DNA with the precision, scale, and price points needed for the AI market.

Gene Pools wrench the cork out of the bottleneck on DNA availability

Fortunately, Twist has spent a decade refining the synthesis platform and in-house expertise that the AI market needs. And we continue to do so. To further empower these AI applications, and to wrench the cork out of the bottleneck on DNA availability, Twist is excited to launch early access for “Gene Pools”. These are unlimited-size pools of gene length DNA. Get as many sequences as you want, up to 1.8kb in length, pooled together in a single tube.

Previously we had launched Oligo Pools and Multiplexed Gene Fragments, which are both made by the pooled synthesis of single oligos up to 500 bases in length. Gene Pools is the logical next step. After synthesizing these highly precise oligos, we use a novel pooled assembly approach to stitch them together, forming the exact gene-length sequences you order. Doing this in a pooled format enables large-scale production, exactly what is needed in the protein screening and discovery space currently being revolutionized by AI. These sequences are then sent in a single tube ready for screening, all in as fast as 10 business days*.

We started working on Gene Pools in 2019, and it was a rocky start in large part because chimera rates were unacceptable. Even when we managed to make pools that met our quality and fidelity criteria, they were overly restricted by stringent design requirements. Today, we hear from the market that the few brave providers promising to make pooled gene length DNA similarly fail on all three criteria.

With Gene Pools, we are proud to have solved the pooled assembly challenge. Through our proprietary pooled assembly platform, our Gene Pools promise zero chimerism, regardless of length, design, or pool size.

You read that right, 0% chimeras.

Additionally, the dropout rate is <1%, and the sequences are provided with uniform representation. Just order as many genes as you need, up to 1.8kb, and get them delivered exactly as you ordered them. We also remove the need for downstream amplification by providing 10 ug of pool material as the standard, thus helping you maintain pool fidelity during cloning.

"I am confident in saying no other available technology provides such unprecedented access to clean, noise-free sequence space in the wet lab."

So, if you need to build a dataset of gene sequences to train or validate an AI model, if you need a precise library of scFvs, if you’re working on precise TCR immunization libraries, or if you’re planning on screening sequences at scales that have long seemed unfeasible, Twist’s Gene Pools are for you. Use them to encode the thousands of exact sequences you are looking to screen, clone them into your vector of choice, and start exploring. I am confident in saying no other available technology provides such unprecedented access to clean, noise-free sequence space in the wet lab.

Though we continue to introduce new and advanced technologies, our pitch is the same: We have parallelized gene synthesis. We can make more high-quality genes in a single run than anyone else, and we make them better than anyone else. Go and use this abundant raw material to improve the world. And as was true more than a decade ago, one question remains: What will you do with unlimited DNA?

Get unprecedented access early by going to our website or connecting with your sales rep.

Gene Pool Specs:

• Sequence length: 300-1800 bp
• Sequence number: Unlimited
• Length variation in the pool: Unbounded
• Sequence similarity within the pool: Unbounded
• TAT: 10 BD#
• Chimera rates: 0%
• Average error rates: 1:3000**
• Dropout rates: <1%†
• Uniformity: 95th/5th of ≤10‡
• Mass: 10 µg as standard. More on request.

*Standard TAT is 10 business days. Pools >50,000 sequences may require up to 15 business days. This timeframe refers to the typical processing and handling time within our facilities before your order is handed over to the shipping carrier. Actual delivery times will vary depending on your location, the chosen shipping method, and the carrier's handling procedures.

** Error rate for the previous year was an average of 1/2911 with a standard deviation of 422.

† Dropout rate refers specifically to accepted sequences. Of fragments accepted in fiscal year 2025, >99% were shipped.

‡ Uniformity metrics are only applicable for sequences within a maximum 15% length variation.

# Gene pools up 50k fragments are shipped within 10BD, up to 150k fragments are shipped within 15BD. This timeframe refers to the typical processing and handling time within our facilities before your order is handed over to the shipping carrier. Actual delivery times will vary depending on your location, the chosen shipping method, and the carrier's handling procedures.