Skip to main content
. 2006 Nov;15(11):2499–2506. doi: 10.1110/ps.062125306

Figure 3.

Figure 3.

Accuracy of prediction Q3 of the Consensus Data Mining method as a function of the upper sequence identity limit. The different curves were obtained by varying the sequence identity thresholds. The sequence identity threshold of 50% gives the best results. At 50% threshold, CDM always performs better than individual FDM applied to the whole sequence (full coverage).