Nature Podcast

In this episode:00:46 An AI that predicts gene interactionsMapping the network of genes that control cellular processes can be difficult to do when gene-expression data is sparse, such as in rare diseases or those affecting tissues that are hard to clinically sample. To overcome this, a team has developed an artificial intelligence system trained on a large, general dataset, and used it to make predictions about gene interactions in data-limited situations. As a test-case they look at the heart condition cardiomyopathy, and show that the system can identify potential interactions that could represent new therapeutic targets.Research article: Theodoris et al.09:08 Research HighlightsMicrobes that can break down persistent ‘forever chemicals’, and why intermolecular distances are the key to keeping gummy sweets chewy.Research Highlight: Microbes take the ‘forever’ out of ‘forever chemicals’Research Highlight: Better gummy sweets are within reach, thanks to physics12:06 Briefing ChatWe discuss some highlights fr