NGS-verified protein variant pools, designed and synthesized for your high-throughput assay.
Protein optimization is a search across a vast functional sequence space. The signal is sparse, and the map is empty. You begin with one sequence — and a sense that there is something better, somewhere out there.
Every mutation from the original candidate is a step into the unknown. Somewhere in that space sits a variant that is faster, more selective, more stable. The direction we have to travel is unknown, and often quite far.
Most mutations will result in either neutral or worse performance. In some cases, there is a valley of low-performing mutations separating our initial candidate from the best-performing variants.
There are many ways that people have done protein optimization. All of which have to make trade-offs based on the realities of preparing physical samples and running experiments.
Historical methods of directed evolution — DMS, SSM, error-prone PCR — are powerful tools for polishing an already good enzyme. They can generate a large number of candidates and leverage high-throughput screening technologies. However, because they stay so close to the original sequence, campaigns often require many rounds to achieve real improvement.
Rational design — by humans or AI — can leap to distant candidates in a single step. But each candidate requires custom synthesis, a significant cost barrier, making high-throughput testing impractical. Most of the space goes untested.
Talaria pairs AI-guided design with in-house DNA synthesis. We ship a single pool with up to five million rationally designed, NGS-verified variants. These variants are designed to cover as much of the search space as possible for your protein, resulting in rich functional data. If we don't find a sufficient candidate in our first attempt, our AI models can use this foundational dataset to refine a design for round 2 and to drill deeply into the most performant versions of your protein.
1K–5M
Variants per pool
120–2500 AA
360–7500 BP coding
3–4 weeks
Design to delivery
NGS verified
Every pool
We chose BM3 hydroxylase as our first benchmark because it’s a hard target — over a thousand amino acids, well studied in industrial biocatalysis, and with a published precedent we could measure ourselves against.
In 2006, a landmark BM3 library defined what rational protein design could do: 6,561 chimeras, assembled by hand. Twenty years later, we built the same kind of library at Talaria — and delivered 3.2 million variants in a single pool.
Tell us about your protein and the throughput of your assay. We’ll respond within two business days.
Or email us directly at contact@talariabio.com
BM3 hydroxylase — 3.2 million variants, NGS-verified.
We’ll send the white paper. You can unsubscribe at any time. See our privacy policy for how we handle your data.
On its way. Check your inbox in a minute.