A new AI model has decoded the DNA pattern of a gene-control sequence called the initiator, allowing researchers to identify it in human genes and predict whether it is present or absent. The sequence appears in about 60% of human genes, according to a study that analyzed roughly 500,000 versions of the initiator.
Genes contain instructions that must be activated at the right time and in the right way for cells to produce the proteins and other molecules they need. One important part of this process involves the initiator, a specific segment of DNA where gene expression begins.
Researchers in the laboratory of University of California San Diego molecular biologist James T. Kadonaga set out to better understand the DNA sequence that makes up this region.
Led by graduate student researcher Torrey Rhyne-Carrigg, the team tested about 500,000 different versions of the initiator using high-throughput DNA sequencing technology. The experiments provided information about how strongly each version was associated with gene expression.
The researchers then used those results to train a machine-learning model. The model learned patterns in the DNA sequences and was able to identify the characteristic sequence pattern of the initiator.
The sequence appears in many human genes
With the initiator’s DNA pattern identified, the researchers searched human genes for the sequence.
They found that about 60% of human genes contain an initiator.
The AI model could also predict whether the initiator was present or absent in human genes. According to Kadonaga, the models provided strong predictions of these two possibilities and decoded the DNA base sequence pattern of the initiator.
The work provides information that can be used to predict the effects of DNA mutations involving the initiator, including mutations associated with disorders.
Using the model to study gene regulation
The data and models from the study could also be used to design synthetic promoters. These are DNA sequences that can control whether genes are turned on or off and can be designed with customized functions.
The researchers see the initiator model as one part of a larger effort to use laboratory experiments and AI to decode information contained in human DNA sequences.
Kadonaga described the human gene expression code as specifying when, where and to what extent genes are turned on or off. The new model focuses specifically on the initiator and represents a small part of that broader code.
The study was published in Genes and Development.
