Talus Bio introduced Ptarmigan-1, an AI model designed to predict where small molecules bind proteins without requiring a 3D protein structure, docking, or a predicted binding pose.
The approach aims to address disordered and flexible proteins— including some transcription factors—that are difficult to analyze with conventional structure-based methods.
Talus says the model places protein residues and chemical compounds in a shared learned representation, using binding data from its MARMOT proteomics platform to train it.
The company reports that Ptarmigan-1 screened 3.4 billion compounds against 20,431 human proteins in under a day and runs about 5,000 times faster than the structure-based comparison it cited.
The company’s results are promising but should be treated as early evidence:
performance can vary by target, and the model is presented as complementary to structure-based tools, which may perform better on well-folded proteins.
Sources:
Welcome to the structure-free future of drug discovery
Talus Bio Launches Ptarmigan-1, the First Structure-Free AI Model for Proteome-Wide Drug Discovery
Structure-free drug discovery | Talus Bio
Talus Bio Launches Ptarmigan-1, the First Structure-Free AI Model for Proteome-Wide Drug Discovery
Talus Bio's Ptarmigan-1 Revolutionizes Small-Molecule ...
Talus Bio presents preclinical data on transcription factor ...
Introducing Ptarmigan-1 for Structure-Free Drug Discovery
Talus Bio's Ptarmigan-1 Breaks Ground in Structure-Free ...
Introducing Drugging the Undruggable, a new blog series ...
Talus Bioscience, Inc. - Drug pipelines, Patents, Clinical trials