The research was conducted at the University of Oregon<\/a> (UO). The tool was developed by computational biologist Andrew Kern and his lab.<\/p>\n Genomes as language<\/p>\n The comparison between DNA<\/a> and written language isn\u2019t just a metaphor. Genomes really are built like text \u2013 a four-letter alphabet of A, T, C and G, combined in different sequences to form genes and chromosomes.\u00a0<\/p>\n What Kern\u2019s lab is most interested in is the misspellings: mutations<\/a>, changes in DNA sequences that accumulate over time and get passed down from generation to generation, leaving a trail that researchers can follow backwards through evolutionary history.<\/p>\n Traditional methods for doing this \u2013 based on math and statistics \u2013 are the gold standard, and in most cases they\u2019re hard to beat. But they\u2019re slow, and they struggle with large or incomplete datasets.\u00a0<\/p>\n A single mosquito chromosome can take hours or even days to decode. That\u2019s a real bottleneck when you\u2019re working at scale.<\/p>\n Borrowing from ChatGPT<\/p>\n To get around this, Kern and his team modified GPT-2 \u2013 the older machine learning architecture that underlies ChatGPT<\/a>.\u00a0<\/p>\n Instead of training it on volumes of English text, they trained it on simulations of genetic evolution across a range of species, including bacteria<\/a>, rodents, mosquitoes, and primates.<\/p>\n \u201cWe can\u2019t repeat evolution, so one of the key workflows we have is developing simulations,\u201d said Kevin Korfmann, lead author of the study. <\/p>\n \u201cThe simulations mimic evolutionary processes, and then we use the outcomes as training data for our deep learning models.\u201d<\/p>\n The model learns to recognise mutation patterns and use them to estimate when two genes last shared a common ancestor<\/a> \u2013 a measure geneticists call \u201ccoalescence time.\u201d Stretches of DNA with many mutations tend to trace back to a distant common ancestor.\u00a0<\/p>\n Those with fewer mutations likely diverged more recently. It\u2019s the same principle that explains why chimpanzees are considered our closest living relatives, while sea sponges \u2013 genetically diverged more than 700 million years ago \u2013 are among the most distant<\/p>\n When the team tested the tool against existing state-of-the-art statistical methods, it performed just as well \u2013 which came as a genuine surprise.\u00a0<\/p>\n \u201cYou never really know what\u2019s going to work when you\u2019re essentially borrowing techniques from a totally different world and applying them to a new problem,\u201d Kern said. \u201cBut this was a case where things worked really well.\u201d<\/p>\n The speed difference, though, was dramatic. Where traditional methods can take hours or days to process a single mosquito chromosome, the new tool does it in minutes.\u00a0<\/p>\n The reason, Korfmann noted, is that the heavy statistical lifting happens during training rather than during each individual analysis.\u00a0<\/p>\n \u201cIt just reads the patterns because all of the expensive statistical work was done up front, during training, which sidesteps the bottleneck,\u201d he said.<\/p>\n The tool also handles incomplete data \u2013 a common headache in genetics research \u2013 without falling apart. For Kern, who regularly works with patchy mosquito genetic databases in his malaria research, that\u2019s not a minor convenience.<\/p>\n Why mosquitoes matter<\/p>\n Insecticides<\/a> have long been one of the main weapons against malaria-spreading mosquitoes. But mosquitoes, like everything else, evolve.\u00a0<\/p>\n Resistance to insecticides is now showing up across mosquito populations worldwide, and understanding how and when that resistance emerged is critical to staying ahead of it.<\/p>\n \u201cA major challenge in preventing the spread of malaria has been understanding the evolution of insecticide resistance,\u201d Kern said.\u00a0<\/p>\n \u201cNow, we can go in with our AI model, ask how long ago these resistance genes arose in the population, and learn about the evolutionary history of this critical carrier of malaria.\u201d<\/p>\n Future research directions <\/p>\n Right now, the model traces ancestry between pairs of genes. The next goal is to scale that up by reconstructing full genealogical trees across multiple lineages simultaneously.\u00a0<\/p>\n Some traditional methods can already do this, but Kern and Korfmann want to get there from a machine learning angle.<\/p>\n \u201cThere\u2019s so much going on in the machine learning field that we haven\u2019t applied yet in our field,\u201d Korfmann said. \u201cThere\u2019s tons of translational work to do to get these novel algorithms working in biology.\u201d<\/p>\n The gap between AI research and biological application, in other words, is still wide. But it\u2019s closing.<\/p>\n The research is published in the journal Proceedings of the National Academy of Sciences<\/a>.<\/p>\n \u2014\u2013<\/p>\n Like what you read? Subscribe to our newsletter<\/a> for engaging articles, exclusive content, and the latest updates.<\/p>\n Check us out on EarthSnap<\/a>, a free app brought to you by Eric Ralls<\/a> and Earth.com<\/a>.<\/p>\n \u2014\u2013<\/p>\n","protected":false},"excerpt":{"rendered":"Scientists have built an AI tool that reads genetic code the way ChatGPT reads text \u2013 scanning DNA…\n","protected":false},"author":2,"featured_media":471728,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[77],"tags":[18,19,17,133],"class_list":{"0":"post-471727","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-science","8":"tag-eire","9":"tag-ie","10":"tag-ireland","11":"tag-science"},"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@ie\/116529228284504563","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts\/471727","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/comments?post=471727"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts\/471727\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/media\/471728"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/media?parent=471727"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/categories?post=471727"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/tags?post=471727"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\n ![\"EarthSnap\"\/](\"https:\/\/www.europesays.com\/ie\/wp-content\/uploads\/2026\/03\/1774817599_888_earthsnap-banner-news.webp.webp\")
\n<\/a><\/p>\n