Researchers used AI to decode the DNA signature of a key genetic ‘on switch’ involved in turning genes on. By analyzing roughly 500,000 DNA sequences, the model identified the initiator in about 60% of human genes, providing a higher-resolution map of how transcription initiation may be controlled. The advance could improve interpretation of harmful mutations by clarifying how sequence changes affect gene activation. It also supports broader efforts to decode the regulatory instructions that determine gene activity patterns. For biotech, better initiator prediction can translate into more precise variant impact assessment in genomics-driven drug discovery and diagnostics.