AI uncovers biological rules to advance medicine Computational biochemist David Hendrix is pioneering the use of artificial intelligence to advance RNA biology. RNA, or ribonucleic acid, is a crucial molecule in all living cells that plays roles in protein synthesis and gene regulation. His research focuses on noncoding RNAs that regulate genes rather than code for proteins. To study these complex molecules, Hendrix’s lab develops advanced AI models and architectures that classify RNA transcripts, analyze protein-coding potential and uncover the biological rules that govern RNA structure and function. By combining deep learning, machine learning and data mining, the team reveals new mechanisms of gene regulation while also validating known biology. A central goal of their work is not only to improve the accuracy of predictions, but to interpret what the models have learned and extract meaningful biological insight. Hendrix’s interdisciplinary expertise has sparked meaningful collaborations across the university, strengthening research at the intersection of computation and life sciences. Among his notable innovations, he was an early leader in applying deep learning to cancer detection using gene expression data, an approach that has gained wide adoption. Exploring the gut-brain axis and deep sea ecosystems Maude David’s research sits at the crossroads of microbiology, neuroscience and artificial intelligence — an intersection that may hold the key to understanding some of the most complex disorders affecting the human brain and unlocking the secrets of deep-sea ecosystems. At the center of her work is the gut-brain axis, a complex, two-way communication network between the gut and the central nervous system. While scientists have long known about this pathway, only in recent years have they begun to understand how trillions of microbes in the human digestive system can influence brain function and behavior. “I am fascinated by the complex relationship we have with our microbiome,” David said. “I work specifically on this pathway where the microbes could potentially modulate sensory cells, that’s two synapses in your brain. So, in a millisecond, the bacteria or their metabolites can ‘touch’ your brain.” Her lab is particularly interested in what role this communication network may play in neurological disorders like autism spectrum disorder (ASD). Using crowdsourced data, David and collaborators discovered that children with ASD have distinct differences in the composition of their gut microbiota compared to their neurotypical siblings. The researchers recruited 111 families that each have two children — one with autism and one without — born within two years of each other and aged two to seven years old. The researchers collected stool samples from the children at three different time points, two weeks data mining → finding patterns, trends and relationships in large datasets Hendrix’s interdisciplinary expertise has sparked meaningful collaborations across OSU. Some of the techniques the Hendrix Lab has used to build a more complete understanding of gene regulation include convolutional neural networks and gated recurrent units. Hendrix has made a GRU model for transcript classification publicly available through his GitHub (@hendrixlab) as part of the mRNN project. 10 OREGON STATE UNIVERSITY / COLLEGE OF SCIENCE
RkJQdWJsaXNoZXIy MTcxMjMwNg==