Table 1.

Examples of assessment of secondary structure predictions using Q3, SOV’99, and SOV_refine (λ = 1)

Index	Sequence	Q3	SOV’99	SOV_refine (λ = 1)
Reference	CHHHHHHHHHHC	–	–	–
Predicted 1	CHCHCHCHCHCC	0.583	0.125	0.149
Predicted 2	CHHHCHHHCHHC	0.833	0.406	0.371
Predicted 3	CHHCCHHHHHCC	0.75	0.523	0.464
Predicted 4	CCCHHHHCCCCC	0.50	0.544	0.459
Predicted 5	CCCHHHHHCCCC	0.583	0.632	0.567
Predicted 6	CCCHHHHHHCCC	0.667	0.806	0.678
Predicted 7	CCCHHHHHHHCC	0.75	0.903	0.797
Predicted 8	CCCHHHHHHHHC	0.833	0.944	0.937

Notice that predicted 3 and 4 indicate another different feature between SOV’99 and SOV_refine. Predicted 3 correctly predicts seven helices (in two segments) while Predicted 4 correctly predicts four helices (in one segment). In this situation, SOV_refine assigns a higher score to predicted 3 as seven correct helices are more consistent to the reference’s 10 helices compared to predicted 4’s four helices. However, SOV’99 in this case assigns a higher score to predicted 4 showing it prefers one segment prediction even though the number of accurately predicted residues is largely different