Utilizing generative AI, MIT chemists efficiently compute three-dimensional configurations of genetic structures
The nuts and bolts of genetics are a sight to behold, even if each cell within your body carries the same genetic sequence. What makes a brain cell unique from a skin cell comes down to the way genes are expressed, and the three-dimensional structure of the genetic material plays a significant role in this.
Now, chemists from MIT have devised a clever way to decipher these three-dimensional structures, slinging generative AI right into the mix. This new technique can whip through thousands of structures in a matter of minutes, quickly outpacing traditional methods for studying the structures.
So, how does this work? Inside the cell nucleus, DNA and proteins form a complex called chromatin. This intricate maze of levels allows cells to pack 2 meters of DNA into a nucleus size of merely one-hundredth of a millimeter. The DNA winds around proteins called histones, forming a structure akin to strings of beads.
Chemical tags called epigenetic modifications can attach to DNA at specific spots, and these tags vary by cell type. They impact the folding of the chromatin and gene accessibility, guiding which genes are expressed in different cell types or at different times within a cell.
In the past 20 years, scientists have devised experimental techniques to determine chromatin structures. One popular one, Hi-C, links together neighboring DNA strands and sequencing them to figure out which segments are near each other. However, Hi-C and similar techniques can be laborious, with data generation from one cell taking around a week.
To break these limitations, MIT researchers have conjured up a model that harnesses the power of generative AI for swift, accurate analysis of DNA sequences to predict chromatin structures in single cells. The AI model they designed can zip through data at whirlwind speeds, shedding light on the otherwise tricky and time-consuming process of predicting chromatin structures.
So, where do we go from here? The researchers have put their model to the test, generating structure predictions for over 2,000 DNA sequences and comparing them to experimental data. The predictions matched up shockingly well, paving the way for a slew of new possibilities in studying the effects of chromatin structures on individual cell functions.
"Our goal was to try to predict the three-dimensional genome structure from the underlying DNA sequence," says Bin Zhang, the study's senior author and an associate professor of chemistry at MIT. "Now that we can do that, it puts this technique on par with the cutting-edge experimental techniques. It can really open up a lot of interesting opportunities."
Greg Schuette and Zhuohan Lao, two sharp MIT grad students, are the paper's lead authors, and the research "appears today in Science Advances."
Keep your eyes peeled for further advancements in this field, as these AI-based methods promise to take our understanding of chromatin organization and gene regulation to new heights. The researchers have also made their data and the model available for others to play with, so strap on your lab coats and get ready to dive in!
- The researchers from MIT have devised a method using generative AI to decode the three-dimensional structures of chromatin in single cells, outpacing traditional methods.
- The new technique, tested on over 2,000 DNA sequences, has shown remarkable accuracy in predicting chromatin structures compared to experimental data.
- In the past, scientists have relied on laborious techniques like Hi-C to study chromatin structures, with data generation from one cell taking around a week.
- Chromatin structures, formed by DNA and proteins within the cell nucleus, are crucial in guiding which genes are expressed in different cell types or at different times.
- The MIT model can swiftly analyze DNA sequences to predict chromatin structures, promising to revolutionize the study of chromatin organization and gene regulation.
- Bin Zhang, an associate professor of chemistry at MIT and the study's senior author, states that the technique now stands alongside cutting-edge experimental techniques.
- Greg Schuette and Zhuohan Lao, sharp MIT graduate students, are the paper's lead authors, and the research has been published in Science Advances.
- The researchers have made their data and model available for others to explore, inviting scientists to delve into the field of AI-based chromatin structure research for potential future advancements in health, biology, and science.
