Figure 2
From: Predicting N-terminal myristoylation sites in plant proteins
Effect of test set pruning on algorithm performance measurements. Accuracy scores (number correctly classified/total number of examples tested) were determined for the following algorithms: HMM80 (closed circles), NMT (open squares), BGM (closed triangles) Expasy (closed squares), and Prosite (open circles). Test set similarity refers to the maximum number of amino acid matches permitted between any two sequences in the set. Number of sequences tested were a) 80 positive, 185 negative (original, unfiltered test set, 96% maximum similarity), b) 63 positive, 128 negative (80% maximum similarity), c) 55 positive, 102 negative (60% maximum similarity), and d) 44 positive, 94 negative (40% maximum similarity).