Fig. 1

A Distribution of cosine similarity scores multiplied by 10 and BLOSUM62 substitution scores from aligned residue pairs in each BAliBASE reference. Each MSA was split into pairwise alignments. A random sample of 100 pairs from the first 5 pairwise alignments from each MSA were extracted so that one MSA did not have too significant of an effect on either distribution, as some MSA’s have much longer sequences or have alignments with much lower pairwise sequence identities than other MSA’s. B Distribution of cosine similarity scores multiplied by 10 and BLOSUM62 substitution scores from random residues in each BAliBASE reference. Four random residues were selected from five random sequences from each MSA and scored against each other. This number of samples was chosen to roughly match the number of samples in (A)