Table 1:

Summary and performance of kataegis detection packages

					Synthetic dataset					WGS dataset
Package	Reference	Available on	Language	Method	Accuracy	nMCC	F1	TPR	TNR	Accuracy	nMCC	F1	TPR	TNR
Katdetectr	[21]	Bioconductor	R	Changepoint analysis (PELT)	0.99	0.98	0.97	0.94	0.99	0.99	0.92	0.83	0.91	0.99
SeqKat	[13]	CRAN	R	Sliding window/exact binomial test	0.84	0.54	0.02	0.93	0.84	0.99	0.85	0.69	0.59	0.99
MafTools	[10]	Bioconductor	R	Sliding window/piecewise constant fit (PCF)	0.74	0.53	0.01	0.96	0.74	0.99	0.85	0.66	0.93	0.99
SigProfilerClusters	[14]	GitHub	Python	Model sample-specific IMD cutoff	0.65	0.52	0.01	0.88	0.65	0.99	0.84	0.68	0.66	0.99
ClusteredMutations	[11]	CRAN	R	Anti-Robinson matrix	0.70	0.53	0.01	0.99	0.74	0.99	0.83	0.61	0.99	0.99
Kataegis	[12]	GitHub	R	Piecewise constant fit (PCF)	0.99	0.80	0.52	0.36	0.99	0.99	0.56	0.03	0.02	0.99

Summary: information of all evaluated kataegis detection packages and their respective performance metrics regarding kataegis classification on 1,024 synthetic samples and 507 a priori labeled whole-genome sequenced (WGS) samples. Accuracy, normalized Matthews correlation coefficient (nMCC), F1-score, true-positive rate (TPR), and true-negative rate (TNR), pruned exact linear time (PELT), piecewise constant fit (PCF), and intermutation distance (IMD).

Note: Highest value per column is underscored.