Solo: doublet identification via semi-supervised deep learning

We developed a semi-supervised deep learning framework for the identification of doublets in scRNA-seq analysis called Solo. To validate our method, we used MULTI-seq, cholesterol modified oligos (CMOs), to experimentally identify doublets in a solid tissue with diverse cell types, mouse kidney, and showed Solo recapitulated experimentally identified doublets. Overall design: To better understand the ability to computationally identify doublets we ran two wells on the 10X v3 platform to generate single cell gene expression profiles and cholesterol-modified oligo count profiles for two mouse kidneys from one mouse. CMO_groups.csv: How CMO barcodes were pooled together. To ensure redundancy during MULTI-seq we used three CMOs per sample for identification. We used the combined counts to demultiplex cells and identify doublets. feature_barcodes.csv: CMO barcodes used and their names.