Table 2.

Methods/studies that map cell lines to tumours.

Reference	Data Input	Multi‐omic integration	Application	Unsupervised/Supervised	Clustering/subtype	Method
						CCL—Tumour integration	Scoring CCL	Selecting CCL
Warren et al (2021) (Celligner)	GE	–	Pan‐cancer	Unsupervised	Subtype	Contrastive PCA + Mutual Nearest Neighbour	Pearson corr. on aligned space	–
						Assignment by k‐NN
Peng et al (2021) (CancerCellNet)	GE	–	Pan‐cancer	Supervised	Subtype	–	Classification score	Multi‐class Random Forest
								“Correct” class: classification score > thr in actual type
Sinha et al (2021) (TumorComparer)	GE, CNA, Mut	Late	Pan‐cancer (independent)	Unsupervised	Subtype	–	Aggregated ranking of weighted Pearson's corr./Jaccard Index	–
Zhang & Kschischo (2021) (MFmap)	GE, CNA, Mut	Intermediate	Pan‐cancer (independent)	Supervised (subtype)	Subtype	ComBat (GE)	Cosine coefficient (latent space)	Neural network classifier on latent space
						Concatenated VAE
Fang et al (2021)	PE	–	Thyroid Carcinoma	Unsupervised	Subtype	–	Pearson's corr.	–
Najgebauer et al (2020) (CELLector)	CNA, Mut, HypMet	Early	Pan‐cancer (independent)	Unsupervised	Clustering	–	Signature length times fraction of samples in group	Eclat clustering
								Map by decision tree
Salvadores et al (2020) (HyperTracker)	GE, HypMet	Late	Pan‐cancer	Supervised	Subtype	ComBat	–	Binomial ridge regression
								“Golden set” from matching data modalities
Batchu et al (2020)	GE	–	Alveolar Rhabdomyosarcoma	Unsupervised	–	–	Spearman's corr.	–
Yu et al (2019) (CompHealth)	GE	–	Pan‐cancer (independent)	Unsupervised	Subtype	ComBat	Spearman's corr.	TCGA‐110‐CL panel: 5 highest score per type
				Supervised (subtype)
Liu, et al (2019a)	GE, CNA, Mut	–	Metastatic Breast Cancer	Unsupervised	Subtype	–	Spearman's corr. (GE and CNA)	–
Ronen et al (2019) (Maui)	GE, CNA, Mut	Intermediate	Colorectal cancer	Unsupervised	Clustering	Multimodal stacked VAE	Euclidean distance (latent space)	K‐means clustering (latent space)
								at least 1 of 5 NN being tumour
Zhao et al (2017)	GE, CNA, Mut	Late	Pan‐cancer (independent)	Unsupervised	Subtype	Distance weighted discrimination	Kendall Rank corr. (GE and CNA)	Similarity in at least 3 out of 4 modalities
							Gene Ontology enrichment score
							Mutation presence
Luebker et al (2017)	CNA, Mut	–	Melanoma	Unsupervised	–	–	Fraction of genome altered	–
							Pearson's corr. (CN)
Vincent & Postovit (2017)	GE	–	Melanoma	Unsupervised	Subtype	–	Pearson's corr.	–
Sinha et al (2017)	GE, CNA, Mut	–	Renal Cancer	Unsupervised	Clustering/Subtype	ComBat	–	Hierarchical clustering (Spearman corr., CN)
				Supervised (subtype)				PAMR classifier (Spearman corr., GE)
Jiang et al (2016)	GE, CNA, Mut, PE	Late	Breast Cancer	Unsupervised	Clustering/Subtype	–	Sum Pearson corr.	Hierarchical clustering (PE, GE)
Sun & Liu (2015)	GE, CNA	Late	Breast Cancer	Unsupervised	Subtype	–	Aggregated ranking of Spearman's corr.	–
Vincent et al (2015)	GE	–	Breast Cancer	Unsupervised	Subtype	‐	Pearson's corr. (group specific)	‐
Chen et al (2015)	GE	–	Hepatocellar Carcinoma	Unsupervised	–	–	Spearman corr.	–
Sadanandam et al (2013)	GE	–	Colorectal cancer	Unsupervised	Clustering	Distance weighted discrimination	–	SAM and PAM for feature extraction
								Consensus‐based NMF
Domcke et al (2013)	GE, CNA, Mut	Late	Ovarian Cancer	Unsupervised	Subtype	‐	sum: CNA Pearson corr. and Mut presence/absence	GE for validation: hierarchical clustering
Virtanen et al (2002)	GE	–	Lung Cancer	Unsupervised	Clustering	Lowess normalisation	–	Hierarchical clustering
								Comparison with known label

CCL, cancer cell line; CNA, copy number alterations; GE, gene expression; HypMetm DNA methylation; Mut, somatic mutations; PE, protein expression.