NVIDIA and Harvard Team Up to Accelerate and Reduce Costs in Genome Analysis Using AI

Scientists from NVIDIA and Harvard have achieved a significant breakthrough in genetic research with the creation of AtacWorks, a deep-learning toolkit designed to streamline rare and single-cell experiments. According to a study published in Nature Communications, AtacWorks dramatically reduces the time and cost of genomic analyses. Tasks that typically require over two days can now be completed in just thirty minutes—thanks to the power of NVIDIA's Tensor Core GPUs.

AtacWorks employs ATAC-seq, a well-established technique used to identify accessible regions of the genome in both healthy and diseased cells. These accessible regions are critical for determining and activating specific cellular functions—such as those in liver, blood, or skin cells. Moreover, they hold vital clues about potential predispositions to diseases like Alzheimer’s, heart disease, or cancer.

Traditionally, ATAC-seq analysis involves tens of thousands of cells; however, AtacWorks achieves comparative results using only a small sample of cells. Researchers successfully applied this toolkit to stem cell datasets, distinguishing DNA segments responsible for red and white blood cell production—subtypes that are challenging to analyze with conventional methods.

The ability to analyze genomes more efficiently not only facilitates the identification of specific mutations and biomarkers linked to diseases but also holds promise for accelerating drug discovery by elucidating disease mechanisms. "With very rare cell types, it’s not possible to study differences in their DNA using existing methods,” explained Avantika Lal, a researcher at NVIDIA and lead author of the study. “AtacWorks can significantly lower the cost of chromatin accessibility data collection and unlock new opportunities in drug discovery and diagnostics.”