Fig. 1

Overview of donor deconvolution and the hadge pipeline. A Schematic example of the cellular components leveraged by single-cell multiplexing experiments. Hashing cell counts and scRNAseq reads with SNP calling by cell are the input to the hadge deconvolution pipeline. B hadge implements 12 methods across two sub-workflows of which seven are hashing-based and five are genotype-based deconvolution that can be run independently, in parallel or jointly, in rescue mode. In rescue mode, the pipeline offers the option to refine hashing results with genotype-based deconvolution methods to rescue failed hashing experiments in the donor-matching process. It compares the concordance in donor identification between hashing and genotype-based methods and identifies the best pair of two strategies based on the calculated Phi score