Table 1.

SPhyR reconstructs a phylogenetic tree for patient CRC1 from (Leung et al., 2017) with larger data likelihood than existing methods

Method	$\log \mathrm{Pr}(D|B,\alpha ,\beta )$	$1\to 0$	$0\to 1$	$?\to 0$	$?\to 1$	# edits	# losses	# par. evo.
SCITE	−447.66	33	54	142	49	278	0	0
SiFit	−471.62	14	96	126	65	301	14	15
SPhyR (k = 0)	−450.70	19	79	138	53	289	0	0
SPhyR (k = 1)	−413.38	13	74	137	54	278	14	0

Note: The input matrix $D=[d_{p,c}]$ has m = 178 taxa (cells) and n = 16 characters (single-nucleotide variants). For each method, we show the data likelihood, the number of $1\to 0$ changes, the number of $0\to 1$ changes, the number of $?\to 0$ changes, the number of $?\to 1$ changes, the total number of changes, the number of losses, and the number of times a character is introduced more than once (parallel evolution).