With nearly 5 million deaths worldwide attributed to antibiotic resistance each year, there's a pressing need for innovative approaches to combat resistant bacterial strains.
Researchers from Stanford Medicine and McMaster University are tackling this challenge using generative artificial intelligence. Their latest creation, SyntheMol, is a cutting-edge model designed to craft structures and chemical formulas for six novel drugs aimed at targeting resistant strains of Acinetobacter baumannii, a leading cause of antibiotic resistance-related fatalities.
Their groundbreaking study, published in the journal Nature Machine Intelligence on March 22, outlines the development of this model and its validation through experimental trials of the newly synthesised compounds.
Dr. James Zou, an associate professor of biomedical data science and co-senior author of the study, underscores the critical need for swift antibiotic development. He explains that their objective was to leverage AI to design entirely novel molecules that have not yet been encountered in nature.
Prior to the advent of generative AI, researchers employed various computational methods for antibiotic development. Their primary approach involved using algorithms to sift through existing drug libraries to identify compounds with the potential to combat specific pathogens. However, this method had limitations in discovering all possible chemical compounds with antibacterial properties.
Generative AI, with its ability to "hallucinate" or generate responses, offers promising avenues for drug discovery. Yet, previous endeavours to generate new drugs using this technology yielded compounds that were unfeasible for real-world synthesis. Consequently, the researchers focused on ensuring that all molecules generated by SyntheMol could be synthesised in a laboratory setting.
The model was trained to devise potential drugs using a library comprising over 130,000 molecular building blocks and a set of validated chemical reactions. It not only generated the final compound but also provided detailed synthesis pathways, offering researchers precise recipes for drug production.
Moreover, the researchers trained the model using existing data on the antibacterial activity of various chemicals against A. baumannii. Armed with this information and its set of building blocks, SyntheMol generated approximately 25,000 potential antibiotics and corresponding synthesis recipes within a mere nine hours. To prevent bacterial resistance, the researchers filtered the generated compounds to include only those distinct from existing ones.
Subsequently, the researchers selected the 70 most promising compounds for further investigation and collaborated with the Ukrainian chemical company Enamine to synthesise them. Enamine successfully synthesised 58 of these compounds, six of which demonstrated efficacy against a resistant strain of A. baumannii in laboratory tests. Furthermore, these new compounds exhibited antibacterial activity against other infectious bacteria prone to antibiotic resistance, such as E. coli, Klebsiella pneumoniae, and MRSA.
To assess the safety of the compounds, the researchers tested two of the six in mice for toxicity. While four compounds did not dissolve in water, the tested compounds appeared to be safe. The next phase involves evaluating the efficacy of these drugs in mice infected with A. baumannii.
Significantly, the six compounds exhibit substantial differences from both each other and existing antibiotics. While the molecular mechanisms underlying their antibacterial properties remain unclear, further exploration of these details could unveil general principles relevant to other antibiotic developments.
Dr. Zou and his team are continuing to refine SyntheMol and extend its applications. Collaborating with other research groups, they aim to utilise the model for drug discovery in heart disease and to develop new fluorescent molecules for laboratory research.
Click here to read the original news story.