AtacWorks, a machine learning toolkit, was detailed today by researchers affiliated with Nvidia and Harvard. The toolkit is designed to bring down the time and cost needed for rare and single-cell experiments. The Nature Communications Journal published a study that showed AtacWorks could run analyses on a whole-genome in just half-hour. Comparatively, the traditional methods take hours.
The Nvidia’s NGC hub of GPU-optimized software has AtacWorks. It has an ATAC-seq method for finding open areas in the genome in cells. It is pioneered by Jason Buenrostro, a Harvard Professor, one of the paper’s co-authors. The ATAC-seq measures the intensity of the signal at the genome’s every spot.
The regions with DNA, such that the fewer cell areas are available to correspond by peaks in signals. The noisier the data appears, the difficult it is to identify the areas of DNA that are accessible. ATAC-seq requires thousands of cells to receive a clean signal. According to co-authors, AtacWorks produce the same quality of results with tens of cells.
AtacWorks was trained on labeled pairs of matching ATAC-seq datasets, one noisy and the other high-quality. It could allow scientists to research with limited cell samples also.