Table 2.

benchmark test results

Feature type	percent identity	query features	homolog features	true positive	false positive	new features
PFAM-A domain	<90%	161	161	160	0	4
	<50%	161	150	149	0	3
PROSITE pattern	<90%	166	165	160	0	5
	<50%	166	153	148	0	4
Uniprot site	<90%	360	305	260	0	64
	<50%	360	288	235	0	56
secondary structure	<90%	1486	1265	1150	1	987
	<50%	1486	1197	1009	1	802
Total	<90%	2283	1896	1730	1	1060
	<50%	2283	1788	1541	1	865

Percent identity indicates the maximum similarity of the sequences in the alignment with the query. 'Query features' is the number of sequence features for the query available in the public databases. 'Homolog features' is the number of features found in the other sequences in the alignment that correspond to a feature in the query. True (or false) positives indicate the number of features propagated by MACSIMS that match (or mismatch) with known query features. 'New features' is the number of features predicted by MACSIMS that are not currently found in the public databases.