Google uses Antigravity to identify an antimicrobial peptide candidate

Google DeepMind described on October 15 how researchers used Antigravity to identify a potential antimicrobial peptide. Its predicted activity concerns pathogens associated with respiratory infections, but laboratory validation is still underway.

The workflow combined searches of the UniProt protein database with ProtNLM, which predicts protein functions from amino acid sequences. Researchers focused on Streptococcus pseudopneumoniae and identified a 97-amino-acid candidate belonging to the Lactococcin 972 family.

Google says Antigravity brings database searches, structural assessment and scientific literature analysis into one working environment. Researchers set the biological objectives and decide which evidence to pursue while the agent coordinates the computational work. The approach addresses the difficulty of screening large numbers of proteins whose functions remain unknown.

The next step is experimental testing of the selected candidate. The announcement describes a computational lead awaiting validation, rather than an established treatment or demonstrated clinical result.