Table 1.

Quality estimation methods of multiple sequence alignment software.

	Structural Benchmark	Simulated Sequences	Commonality-Based
Scalability	Low	High	High
Pre-Built Alignment	Yes	Yes	No
Scoring Methods	Sum of pair score and true column score	Sum of pair score and true column score	Multiple overlap score and head-or-tail score
Dependency	Protein structure	Probabilistic model	/
Test Sets	Fixed	Configurable	Not limited
Drawbacks	Limited to the diversity of benchmarks	Adopted model may have defects	Tested software can make common mistakes
Examples	BAliBASE [91] Pfam [95] HOMSTRAD [96]	ROSE [97] INDELible [98] Dawg [99]	MUMSA [100]