Bioinformatics Seminar
Time: 11AM
Venue: Zoom
20 September 2022
Demultiplexing samples and identification of doublets using single cell hashing data
Jianan WangWEHI Bioinformatics
Cell hashing technology enables labelling the cells from different samples via staining droplets with sample specific hashtag antibodies, which allows for loading numerous cells for sequencing. Hashing can reduce the sequencing cost, increase the experimental efficiency, and avoid batch effects across different samples. However, the hashing technology can also produce cross-sample doublets, which will influence the reliability of the analysis results. Here, we propose a computational method based on Mahalanobis distance utilizing the cell hashing data. We demonstrate our method can accurately assign a cell to its original sample and robustly detect doublets. Using the two in-house hashing datasets, we also show that our method works well for both high-quality and low-quality data, and significantly improves the sample assignment efficiency. Our approach will improve the scRNA-seq data quality and facilitate the downstream analyses.