A New Method for Automated Cell Group Classification from Single Cell RNA Sequencing (scRNAseq)

Tech ID: 31841 / UC Case 2019-153-0

Background

Genomics (transcriptome, epigenome, genome, etc.) conveys the most comprehensive information of biological systems and cellular entities. Therefore, it is being increasingly used in research and clinics to classify cells from various developmental origin and functional background to aid scientific discover and medical practice. Especially at single cell level, genomic information has the potential to impact treatment options and medical outcome. However, classifying cells by current methods involves a lengthy bioinformatic analysis procedure that requires expertise not only in biology and medicine but in computer science as well, making it a daunting task for many researchers and clinicians, despite the tremendous healthcare value that single cell genomics provides. Thus, a simple and universally applicable approach will facilitate precision medicine and scientific research.

Technology Description

Researchers at UC San Diego have developed a method to classify cells based on genomic information using deep neural network (dNN), as well as the accompanying data storage structure and user interface. It takes user provided single cell genomic data, automatically identifies each and every cell within the dataset and returns the information back to the user. The inventors propose a single cell genomics classification approach, including backend dNN based analytic algorithm and user-friendly software interface to classify any individual cell without the need of manual analysis at the user end. It not only greatly simplifies the process of cell type identification, allowing anyone or any organization to instantaneously obtain critical information regarding biological sample of interest, but also provides unprecedented scalable and reliable classification of any sample size, from one cell to infinite number of cells.

Applications

As precision medicine advances, it is increasingly pressing to accurately identify the specific cells that are preferentially affected by disease state or responsive to treatment. For example, by combining our method with drug discovery pipelines, a cancer drug development company can easily identify the specific cancer cells within a tumor that are responsive or resistant to therapy. In turn, this knowledge will help the company to design drug delivery system that are tailored for the said cell type and improve the efficiency of drug therapy. Along the same line, our method can be combined with gene therapy to identify cells carrying different somatic mutations, which may expose them to certain disorder.

Advantages

This methodology involves a single cell genomics classification approach, including backend dNN based analytic algorithm and user-friendly software interface to classify any individual cell without the need of manual analysis at the user end. It not only greatly simplifies the process of cell type identification, allowing anyone or any organization to instantaneously obtain critical information regarding biological sample of interest, but also provides unprecedented scalable and reliable classification of any sample size, from one cell to infinite number of cells.

State Of Development

The state of development is in the experimental stage. The inventors have preliminary data suggesting that the method is capable of classifying different types of cells from different sources. The model is based on thousands of cells, covering over 20 types of cells from the human brain. In the future, the researchers are planning to strengthen the model by continuously expanding training dataset.

Intellectual Property Info

The technology is patent pending and available for licensing.

Patent Status

Patent Pending

Contact

Learn About UC TechAlerts - Save Searches and receive new technology matches

Other Information

Keywords

Cell classifier, machine learning, single cell RNAseq, precision medicine, single cell genomics, deep neural network

Categorized As