Breakthrough AI tool uncovers ‘dark side’ of the human genome

As the advances in the field of artificial intelligence (AI) continue, a breakthrough AI tool has managed to explore overlooked regions in human DNA while looking for microproteins that may play roles in disease.

Indeed, scientists at the Salk Institute for Biological Studies in San Diego, California, have developed ShortStop, a machine learning framework to help them study a smaller subclass of proteins that have long been in the shadow of their larger counterparts, per a report on July 31.

What the breakthrough AI tool does

Now, they can analyze the mysterious dark side of the genome in search of these microproteins as they probe genetic databases and identify DNA stretches in the genome that likely code for microproteins, according to the findings published in BMC Methods.

Furthermore, ShortStop can also predict which microproteins are most likely to be biologically relevant, which saves precious resources in terms of time and finances required in the search for microproteins involved in health and disease mechanisms. The new AI tool does this by examining the existing datasets and pinpointing previously untrackable microproteins. 

As it happens, the researchers have already tested the tool in analyzing a lung cancer dataset to identify 210 completely new microprotein candidates, narrowing down to one standout validated microprotein, potentially creating useful therapeutic targets in the future.

According to senior author Alan Saghatelian, professor and holder of the Dr. Frederik Paulsen Chair at Salk, recent discoveries have indicated thousands of small, hidden, and generally overlooked proteins with fewer than 150 amino acids, making them harder to detect using standard methods:

“For a long time, scientists only really studied the regions of DNA that coded for large proteins and dismissed the rest as ‘junk DNA,’ but we’re now learning that these other regions are actually very important, and the microproteins they produce could play critical roles in regulating health and disease.”

How it works

Specifically, ShortStop optimizes the discovery of certain DNA stretches called small open reading frames (smORFs), which can contain the instructions for making microproteins, by sorting microproteins into functional and nonfunctional categories, thanks to its training as a machine learning system.

Notably, this training procedure relies on a negative control dataset of computer-generated random smORFs. The AI tool compares discovered smORFs against these decoys to quickly decide whether a new smORF is likely to be functional or nonfunctional.

Although ShortStop can’t say with total accuracy whether a smORF will code for a biologically relevant microprotein, the two-class system significantly narrows down the experimental pool, reducing the time researchers spend on manually sorting through datasets and failing at the bench.