Table 1.

Comparison of donor deconvolution pipelines

	Demuxafy	cellHashR	HTOreader	hadge
Framework	Singularity	R	R	Nextflow
Available genotype-based methods	5	-	Souporcell	5
Available hashing-based methods	-	7	HTOreader	7
Doublet detection methods	7	-	-	1
Concatenating	-	-	-(*)	+
Parallelized	-	-	-	+
Pre-processing tools	Samtools (*)	ProcessCountMatrix, PlotNormalizationQC	HTOClassification	Samtools
Variant calling tools	Freebayes(*), cellSNP-lite (*)	(not relevant for hashing-based)	(not relevant for hashing-based)	Freebayes, cellSNP-lite
Associating clusters and donors	Only through reference SNP genotypes	(not relevant for hashing-based)	From hashtags to donors based on confusion matrix	From hashtags to donors based on matching Phi score
Combining results	+	+	-	+
scverse compatibility	-	-	-	+

The “Pre-processing tools” and “variant calling tools” columns specify the respectively used tools that are (optionally) run before the deconvolution tools. “Concatenating” refers to the functionality to concatenate hashing-based and genotype-based deconvolution methods. “Combining Results” refers to the functionality that allows the merging of results from multiple methods into a single data frame during a single run, based on the users’ choice. ( +) The pipeline supports the mentioned functionality. ( −) The pipeline does not support the mentioned functionality. (*) The software is required as part of additional preprocessing outside of the pipeline