IT News on October 16th: The issue of antibiotic resistance is intensifying, and there is an urgent global need for new types of antibiotics. However, the development of new drugs is often slow and cumbersome. Google has launched the Google Antigravity intelligent development platform, which can effectively simplify this process.

Google Antigravity provides an intelligent scientific workstation that supports biologists in quickly discovering, structurally evaluating, and screening potential targets, integrating what would otherwise require dispersed searches across multiple databases into a single session. The platform also utilizes the ProtNLM model to help researchers scan millions of uncharacterized proteins in the UniProt database; this model can directly predict biological functions from amino acid sequences.

Researchers have discovered a new type of antibacterial peptide candidate molecule through Google Antigravity. It is predicted that this molecule is active against pathogens causing respiratory infections, and currently, this candidate molecule is undergoing wet laboratory validation.

As cited in a blog post from IT, antimicrobial peptides, especially bacteriocins, are very promising alternatives to traditional antibiotics. These substances have naturally evolved the ability to target fundamental weaknesses of bacteria, making it difficult for pathogens to develop resistance. Currently, billions of natural amino acid sequences have been accumulated in global databases, but the functions of a large number of these sequences remain unmarked and are still in a state of "dark matter."

The ProtNLM deep learning model developed by Google can directly read the evolutionary "grammar" of amino acid sequences and predict biological functions from the original protein sequences, adding high-confidence functional annotations to a large number of UniProt entries.

On the Antigravity platform, researchers collaborate with AI agents to conduct studies: researchers set biological objectives and control the research strategy, while the agents are responsible for coordinating across databases, conducting structural analyses, and synthesizing literature, among other intensive tasks. This model compresses what would otherwise require days of separate database screenings into a single collaborative session, significantly lowering the barriers to biological research.

The platform also comes with a Biomolecular Structure Viewer 3D visualization tool, allowing researchers and agents to collaborate on a shared 3D canvas. Agents can retrieve structures from a AlphaFold database to dynamically drive the visualization and highlight key functional areas; researchers can also mark areas of interest in 3D space, ask questions to the agents, and conduct interactive research based on these structures.

In the past, such research and development work relied on dispersed specialized software. However, Google Antigravity integrates the entire scientific reasoning process into one place, helping researchers to more efficiently explore the vast biological search space and filter out candidates with high confidence levels. This allows scientists to devote more time to verifying high-quality hypotheses and conducting experimental research.