Understanding Gene Regulation through Machine Learning

2019-12-29

In this era of big data, artificial intelligence has become an ally for scientists. For example, machine learning algorithms are helping biologists understand molecular signals that control gene function. Given the significant advances in genetics, it is easy to assume that scientists in the 21st century have mastered a clear and fast method for genome sequencing and identifying which genes can be expressed and which cannot. Gene expression is the process by which information encoded within genes produces key products, such as proteins. However, as new algorithms are developed to analyze more data, they become more complex and difficult to interpret. Biologists Justin B. Kinney and Ammar Tareen have designed advanced machine learning algorithms to make it easier for biologists to understand. Their algorithm is an artificial neural network (ANN). Inspired by the way neurons connect and branch in the brain, artificial neural networks are the computational foundation of advanced machine learning. Despite the name, they are not specifically designed for studying the brain. Tareen and Kinney both use artificial neural networks to analyze data from the massively parallel reporter assay (MPRA) experimental method. Unfortunately, the way standard artificial neural networks are shaped based on MPRA data is very different from how scientists pose questions in the life sciences. This mismatch means that biologists find it difficult to explain how gene regulation occurs. Now, they have developed a new method that bridges the gap between computational tools and biologists' ideas. They have created a custom artificial neural network that mathematically reflects common concepts in biology regarding genes and the molecules that control them. In this way, machine learning algorithms are forced to process data in a way that biologists can understand. Using this data, biologists can create artificial neural networks to predict which molecules control specific genes in a process called gene regulation. Cells do not need all proteins. Instead, they rely on complex molecular mechanisms to turn on or off the production of proteins as needed. Diseases often occur when these regulations fail. Kinney says, "Understanding the principles of gene regulation is often a prerequisite for developing molecular therapies for diseases." These efforts highlight how modern industrial AI technology can be optimized for the life sciences. Kinney's laboratory has validated this new strategy for creating custom ANNs and is now applying it to investigate various biological systems, including key genes related to human diseases.

RECENT AI NEWS

RECENT AI TOOLS