Marker Genes Can Be Substituted By Neural Networks In RNA Sequencing
A novel and efficient approach have been discovered by scientists at Carnegie Mellon University to differentiate cell types after scRNA-seq (single-cell RNA sequencing). The new technique entails the use of neural networks and controlled machine learning techniques instead of marker genes—that aren’t accessible for all cell types.
Utilizing this novel, automated method, scientists can examine all scRNA-seq data and pick merely the parameters that are required to differentiate one cell type from another. This offers scientists with the capability to examine and compare all cell types. The study authors also explain a web server known as scQuery that allows the method to be used by any scientist.
The new technique will be implemented as a component of the new Human BioMolecular Atlas Program of National Institutes of Health, which is developing the human body’s 3D map, which will demonstrate how tissues vary on a cellular level.
To validate the model, scRNA-seq data were used by the scientists from an animal study of a condition resembling Alzheimer’s. As estimated, identical numbers of brain cells were observed in both diseased and healthy tissue, with the unhealthy tissue having notably more immune cells as a retort to disease.
The scientists utilized their automated pipeline and techniques to develop the scQuery web server that speeds up new scRNA-seq data’s comparative analysis. After a single-cell experiment is processed into scQuery, the team’s matching methods and neural network quickly spot correlated subtypes of cells, as well as earlier studies of identical cells.
Similarly, a synthetic molecule has been developed by another research team from Carnegie Mellon University that can identify and attach to double-stranded RNA or DNA under standard physiological settings. The molecule can offer a new platform for bringing up techniques for the analysis and treatment of the genetic conditions. Their results are issued in Communications Chemistry.