Artificial Intelligence, BioPharma, Pharma

Gritstone Oncology showcases AI-based platform’s neoantigen-discovery chops

The company published data in Nature Biotechnology comparing the ability of its EDGE platform to detect neoantigens to those of industry-standard methods.

A company developing immunotherapies in oncology has published data that it said demonstrates the ability of its artificial intelligence-based drug discovery platform to find therapeutic targets.

Emeryville, California-based Gritstone Oncology said Monday that it had published the data in the journal Nature Biotechnology about the platform, called EDGE, a trademarked acronym for Epitope Discovery in cancer Genomes.

The company used three datasets to validate EDGE as a machine learning-based system for identifying neoantigens, which are cancer cell proteins to which the immune system has not been previously exposed.

“Neoantigens are critical targets of immunotherapy and can drive an effective anti-tumor T-cell response, yet existing tools have had low success rates in identifying true neoantigens,” Gritstone Chief Technology Officer Roman Yelensky said in a statement.

The authors first showed that the platform is able to predict the presence of proteins on cancer cells with up to nine times greater accuracy than industry-standard methods. Second, they used neoantigens that had been independently validated and published – most of them from the National Cancer Institute – to compare the performance of EDGE versus other prediction tools. In doing so, the researchers used EDGE to identify validated tumor-specific neoantigens for 11 of 12 patients, while standard approaches identified only four of 12. And third, the company used routine blood samples from nine patients with non-small cell lung cancer who were receiving PD-1 or PD-L1 checkpoint inhibitors to find that EDGE enabled identification of circulating T cells for true neoantigens in five of them.

Gritstone is one of numerous companies that have launched in recent years that employ artificial intelligence and machine learning to aid drug discovery and development. AI and ML’s use in this fashion is based on the premise that they can identify potential targets more quickly and accurately than traditional methods like X-ray crystallography. In July, for example, Cambridge, UK-based Healx raised $10 million in a Series A funding round to use AI in drug discovery and development. That same month, San Francisco-based Verge Genomics raised $32 million for the same purpose. Another company, Salt Lake City-based Recursion, got Food and Drug Administration authorization to start a clinical trial of a drug it discovered via machine learning.

Large drugmakers have been exploring the field’s potential as well. Earlier this year, researchers working under a partnership between Swiss drugmaker Novartis and chip maker Intel used deep neural networks to cut the time for analyzing microscopic images from 11 hours to 31 minutes.

Photo: MF3d, Getty Images

Shares1
Shares1