Clemson University researchers pave the way for artificial intelligence precision medicine

April 12, 2024April 12, 2024

Two people take the same medication to treat similar illnesses. One patient recovered quickly, while the other realized there was no real benefit from the treatment.

Why the same drug doesn’t always produce the same results in different patients is part of what Clemson University researchers are trying to discover in the field of precision medicine.

Zhana Duren, an assistant professor in the Department of Genetics and Biochemistry, is delving deeper into humans’ genetic makeup to unlock answers to this and other questions. He co-authored a paper on this topic with postdoc Yuan Qiuyue. Duren and Yuan both work at Clemson University’s Center for Human Genetics in Greenwood, South Carolina.

new method

Researchers are using a novel approach to better understand the workings of gene regulatory networks (GRNs) by applying two relatively new tools, big data and artificial intelligence, like road maps, showing genes, proteins and other How matter interacts uniquely between people.

Duren explained that GRN maps the complex interactions between genes, regulatory elements and proteins and is key to understanding how genetic variation affects phenotypes such as drug response. Each person has a unique GRN shaped by their specific genotype, which explains why the same drug can cause different reactions in different people.

To explain individual genetic variation in the context of unique GRNs, the goal is to answer key questions, such as how and why genetic variation affects individual phenotypes through complex GRN interactions, Duren said. By elucidating these mechanisms, we pave the way for predicting drug responses based on individual genetics, allowing the development of more targeted therapies and minimizing ineffective treatments.

The problem for researchers, Duren said, is that most disease-related genetic variations are hidden in regions of our DNA that don’t directly code for proteins. This makes it difficult to understand how they affect our health.

Turning to artificial intelligence

To help solve this puzzle, Duren and Yuan turned to artificial intelligence and big data analytics. They developed the LINGER lifelong neural network for gene regulation, a novel deep learning-based method that can infer GRN from other cell-level data.

With the help of new tools, Duren and Yuan’s findings promise to more accurately predict how GRNs work.

Duren points out that many methods for gene regulatory network inference have been developed over the past two decades. However, our system benchmarks based on experimental data show that these existing methods are approximately 17% to 29% more accurate than random predictors. The new method improves this to 125% above random predictions, a relative increase of four to seven times.

He added that because this is a major advance in basic research, it will have the potential to lead to discoveries in a wide range of biomedical research areas.

The results reported by the duo are not without challenges. Chief among them is the sparseness of data.

Because it’s single-cell data, the number of observations we get on each cell is very limited, Duren said. Gene regulatory networks are a very complex problem that require large amounts of data to learn. But independent data are available. We have data from many single cells, but they are not independent and are not sufficient for this task.

Potential applications

Duren said the research has potential applications in many fields, including molecular biology, developmental biology and medical health research. Durham also noted that these studies have the potential to improve understanding of drug addiction, which could lead to the development of more effective treatments.

We are currently applying this approach to the field of drug treatment. I have three collaborative projects working on this; one is applying it to cocaine addiction.

The team’s research is made possible in part by two grants of $2.2 million from the National Institutes of Health in 2023 and 2024.

The paper “Inferring Gene Regulatory Networks from Single Cell Multi-Group Data Using Atlas-Scale External Data” was published by the top peer-reviewed journal “Nature Biotechnology”.