Figure 3
From: Improving pairwise sequence alignment accuracy using near-optimal protein sequence alignments
Comparison of near-optimal sequence alignments to structural alignments. (A) Sampled sequence alignments within 95% of the optimal score (for Zuker alignments), or the top 100 alignments (for probA); (B) sampled sequence alignments within 75% of the optimal score (for Zuker alignments), or the top 500 alignments (for probA). The X-axis has been divided into three groupings, representing three levels of sequence similarity: high (E() < 10-10), medium (E() < 10-5), and low (E() < 0.02). Within the similarity groups, different combinations of scoring matrices and gap penalties are shown. The Y-axis reports the fraction of protein families with a near-optimal sequence alignment that has a structural similarity score better than the best structural alignment (blue), the median structural alignment (grey), and the lowest structural alignment score (black). Structural alignments were not used if the number of aligned residues was less than 50% of the maximum number of aligned residues for that pair.