Table 1.

Overall performance of the tested methods

Criterion	EMMA	WITCH-ng-add	MAFFT--add	MAFFT-linsi--add
Scalability	All	All	All but largest	$\le 1000$ sequences
SPFN-random	1	2	3	1
SPFN-clade	2	3	4	1
Speed	2	2	1	3

For scalability, we indicate which study datasets the method completed on

For SPFN and speed, we indicate the relative position (1 is best, 2 is second best, etc.)