Scientists at Stowers have made a breakthrough in understanding how AI models learn from DNA. They have developed a new method to visualize this process, which is called PISA.

This method allows researchers to see what genomic AI models are learning, and to distinguish between actual biological findings and experimental bias.

Further details about this development and its potential applications, including its use in training more focused models and uncovering insights into gene regulation and genetic disease, are available from the source.