Talus Bioscience released Ptarmigan-1 this week, a structure-free AI model the company says predicts small molecule binding sites across the entire human proteome — including the roughly 40 percent of human proteins that lack stable 3D structures and fall outside the reach of conventional structural biology and tools like AlphaFold.

The Seattle-based company, which has raised $28 million in venture and non-dilutive funding to date, trained the model on at least five years of proprietary data generated by its MARMOT platform (Measuring Modulation of Transcription), a label-free proteomics assay that captures snapshots of proteins inside cells without removing or tagging them.

According to internal company benchmarks, Ptarmigan-1 runs 5,000 times faster than structure-based methods because it skips the protein folding step entirely. Talus claims it can screen more than three billion compounds against the human proteome in a single day. These figures are vendor benchmarks, not independently verified results, and R&D managers evaluating the tool should weigh them accordingly.

Proof of concept in STAT6

Talus has shared results from a proof-of-concept study in which the team used Ptarmigan-1 to evaluate a STAT6 inhibitor series — a validated inflammatory disease target with a binding site that resists structural modeling. The company reports that the model outperformed structure-based methods in selecting successful drug candidates, despite never having seen the target or the molecules during training. The model also flagged small molecules binding to STAT6's flexible pocket, and third-party labs validated those candidates. The company has not disclosed hit rates or sample sizes beyond this single series.

Technical details appear in a preprint posted on bioRxiv describing the architecture and training data. Rather than forcing proteins into a 3D structural space, Ptarmigan embeds proteins and compounds in a shared high-dimensional mathematical space and identifies candidate compounds by proximity within it.

"The data we're building at Talus is structure-agnostic, meaning we can measure proteins whether or not they hold a fixed shape," said co-founder and CTO Lindsay Pino, PhD. "That means the model can learn just as well from flexible or intrinsically disordered proteins as it does from structured ones, which is what lets it generalize to targets nobody's had a way to study before."

A different measurement strategy

The company's founders, Pino and CEO Alex Federation, PhD, met in Seattle about a decade ago and collaborated on the technology now underpinning the platform. Federation trained in the laboratory of Jay Bradner, MD, then at Dana-Farber Cancer Institute and now at Amgen, where the lab worked on molecules binding genomic targets, several of which reached clinical use.

"My big excitement when I was starting my independent career was trying to ask the question, 'could these new technologies actually help us unlock these undruggable targets?'" Federation told GEN. "There is a lot of interest in this problem."

The platform's core measurement addresses a practical lab limitation: unstructured proteins fall apart outside the cell, making them hard to purify or analyze in standard workflows. Talus's assay avoids labels because tags alter protein dynamics.

"Our platform is essentially a unique way to actually watch what these proteins are doing in the cell in their native state without having to take them out of the cell and letting them fold and function within that native environment where all their partners are," Federation said. "That gives us really for the first time the ability to see them in their native state and find molecules that can interact with these proteins in their native state."

He described the approach as "like we're taking a cell and taking snapshots of where the proteins are at any given time and what molecules are sticking to those proteins at any given time."

Access and business model

Ptarmigan-1 is free to scientists through a public portal, with a limited number of experiments available at no cost. Larger target-specific campaigns require direct collaboration — for example, with a company that has a validated target and an existing assay but lacks viable molecules. Federation said collaborations would enable "search screens on the order of billions of compounds" and expressed interest in partnering on future model iterations.

Talus also runs an internal pipeline. Its most advanced molecule targets brachyury, encoded by the TBXT gene, in chordoma — a rare bone cancer. Brachyury is typically expressed only in embryonic stem cells, and the cancer reactivates it.

The free-access tier will be maintained long term, the company says, while commercial interest spans oncology, immunology, neurology, and cardiometabolic disease. Federation frames Ptarmigan-1 as "a first tool out there that can really broaden the landscape … beyond just those structure targets," inviting researchers to apply it to the proteins they care about.