TABLE 2.

Comparison of databases for the identification of a abscess isolate that seemed to be a Streptococcus sp. by routine phenotypic methods but was identified by 16S rRNA gene sequence homology as Gemella bergen^a

Factors to compare between databases	Database
	GenBank	MicroSeq	RDP-II	RIDOM	Our internal database
Total no. of sequences in databasea	About 20,000,000, with about 90,000 16S rRNA gene	About 1,400	About 6,000	More than 300	About 1,500 clinical strains
Was a similar sequence in the database?	Yes, several; closest were G. bergeri (Y13365) and an unnamed oral strain	No, but Gemella haemolysans was identified as the closest relative at 6.5%; thus, one must compare to other databases also	Yes	No	Yes; 0% difference from isolate from finger abscess
Can sequences be imported from another database?		Yes; import best selected sequences from GenBank
Software comparison of our isolate with G. bergeri (Y13365)	0% difference, 100% homology to G. bergeri (Y13365)	0% difference from imported G. bergeri (Y13365)	0.963 related to G. bergeri (Y13365) in database	NA
Comparison with a second tier of closely related strains	There are two sequences deposited as G. haemolysans ATCC 10379; one with GenBank no. L14326 is good quality, one with GenBank no. M58799 has too many N's; L14326 is 93% similar	G. morbillorum and G. haemolysans are equally related at 6.52% dissimilarity	G. haemolysans ATCC 10379 is related at 0.677; these sequences are generally imported from GenBank; the poor-quality sequence, M58799, looks as if it is more closely related, with a 0.739 relative relatedness	NA

^aFor all the public databases, total numbers of sequences are increasing rapidly.