Alignment and clustering of phylogenetic markers - implications for microbial diversity studies

BMC Bioinformatics

Table 3 Top standard methodologies and performance of VI-cut.

MSA	Clustering	Distance	mean VI
MUSCLE	VI-cut	adaptive	0.0589
ClustalW	VI-cut	adaptive	0.0595
ClustalW	fn	0.03	0.0688
MUSCLE	fn	0.04	0.0691
ClustalW	fn	0.04	0.0697
MUSCLE	fn	0.05	0.0748
NAST	VI-cut	adaptive	0.0762
ClustalW	fn	0.02	0.0838
NAST	fn	0.05	0.0845
MUSCLE	fn	0.03	0.0860
NAST	fn	0.06	0.0872
ClustalW	fn	0.05	0.0942
NAST	fn	0.04	0.0992
MUSCLE	fn	0.06	0.1025
ClustalW	fn	0.06	0.1176
NAST	fn	0.03	0.1222
MUSCLE	fn	0.02	0.1370
ClustalW	fn	0.01	0.1505
NAST	fn	0.02	0.1633
NAST	fn	0.01	0.2362
MUSCLE	fn	0.01	0.2629

Methods are ranked by their mean VI-distance over 10 simulated datasets. We constrained the results to commonly accepted methods using furthest neighbor clustering and distance thresholds less than 0.07. A distance threshold of 0.01 is consistently among the worst performing methodologies. VI-cut consistently results in the best clustering for each MSA.

ISSN: 1471-2105