Table 1. Algorithms used to quantify cell identities.

Algorithm	Data input	Molecular identifiers	Basis for selection of molecular identifiersa	Quantification	Single-cell source	Dimension reduction	Example application
CellNet, 2014 [20]	RNA-seq data	Key gene regulatory networks from database	Specificity, regulatory influence	Similarity in the gene regulatory network expression distribution	Mouse, human	N/A	Identification of transcription factors for transdifferentiation; quantification of cell identity score
Mogrify, 2016 [25]	Cell types	Differentially expressed genes; core transcription factors from database	Abundance, specificity, regulatory influence	Scores of key transcription factors	Mouse, human	N/A (MDS used in later steps for landscape construction)	Identification of transcription factors for transdifferentiation; prediction of transdifferentiation potential of a cell type
sci-MET, 2018 [17]	Single-cell bisulfite sequencing data	Top 1000 variable methylation regulatory loci from database	Variance	Correlation in DNA methylation pattern between sample data and database	Human	NMF followed by t-SNE	Prediction of cell identity

Abbreviations: multidimensional scaling (MDS), non-negative matrix factorization (NMF), t-distributed stochastic neighbor embedding (t-SNE)

^aAlthough abundance is not always explicitly mentioned as a selection criterion, it is an implicit criterion for all algorithms due to sensitivity limits in experimental measurements.