Machine learning aids in the discovery of the human core promoter element

–Anuska Sen, Team bioXone

Transcription of genes by the TATA-box is a very common subject among biologists. However, it is only 25% of the human genes that are transcribed by the TATA box. The mechanism behind the rest 75% of genes was thought to be a mystery until researcher Long Vo Ngoc and group identified a core promoter element named DPR using machine learning.

A large number of such DPR (downstream core promoter region) variants were created by a method devised by the researchers, which is known as HARPE (high-throughput analysis of randomized promoter elements). The researchers have reported that these results prove that in humans, DPR is a significant core-promoter element. Another interesting finding of the study is that majority of the promoters show the presence of either DRP or TATA box elements and not both together. This further validates the fact that DRP is responsible for the transcription of a large chunk of genes in the human genome.

DPR does not have any clear sequence pattern. “There is hidden information that is encrypted in the DNA sequence that makes it an active DPR element” as said by Kadonanga, a researcher in the team. The importance of machine learning in this study lies in the fact that it can decipher this code, which could not be made possible by humans with lab-based experiments.

This article has been published in Nature

Source: Vo Ngoc L., Huang C.Y., Cassidy C.J., Medrano C., Kadonaga J.T.; (2020).  Identification of the human DPR core promoter element using machine learning; Nature.