Skip to main content

Table 6 The effect of parameters used for clustering on accuracy of SCRs. Evaluation was performed on families of the Master data set which have more than two domains (the set of multiple familes). Acc SCR (SD): Accuracy when clustered by SD score. Acc SCR (NAS): Accuracy when clustered by Normalised Alignment Score [14]. Acc SCR (PID): Accuracy when clustered by percentage identity. Difference(SD-PID): Difference in SCR accuracy between clustering on SD and PID. p : Wilcoxon Signed Rank test probability (SD-PID).

From: OXBench: A benchmark for evaluation of protein multiple sequence alignment accuracy

PID Average Number of Families Acc SCR (SD) Acc SCR (NAS) Acc SCR (PID) Difference (SD – PID) p
0–10 6 25.9 24.2 24.0 1.9 0.584
10–20 22 57.0 54.8 50.0 7.0 0.00604
20–30 42 76.8 76.7 75.9 0.9 0.379
30–50 130 91.3 91.1 90.9 0.4 0.0163
50–100 199 98.5 98.5 98.4 0.1 0.969
Total 399 90.5 90.3 89.8 0.7 0.000559