Machine learning dissection of human accelerated regions in primate neurodevelopment

• DOI: 10.1016/j.neuron.2022.12.026
• Publication Date: 2023-01-13T15:52:25Z
• Authors: Sean Whalen; Fumitaka Inoue; Hane Ryu; Tyler Fair; Eirene Markenscof...; Kathleen Keough; Martin Kircher; Beth Martin; Beatriz Alvarado; Orry Elor; Dianne Laboy Cintron; Alex Williams; Md. Abul Hassan S...; Sean Thomas; Robert Krencik; Erik M. Ullian; Arnold Kriegstein; John L. Rubenstein; Jay Shendure; Alex A. Pollen; Nadav Ahituv; Katherine S. Pollard
• Keywords: Exaptation

Using machine learning (ML), we interrogated the function of all human-chimpanzee variants in 2,645 human accelerated regions (HARs), finding 43% HARs have with large opposing effects on chromatin state and 14% neurodevelopmental enhancer activity. This pattern, consistent compensatory evolution, was confirmed using massively parallel reporter assays chimpanzee neural progenitor cells. The species-specific activity accurately predicted from presence absence transcription factor footprints each species. Despite these striking cis effects, a given HAR sequence nearly identical suggests that did not evolve to compensate for changes trans environment but instead altered their ability bind factors present both Thus, ML prioritized functional neurodevelopment revealed an unexpected reason why may evolved so rapidly.