Figure S1 - Adaptive GDDA-BLAST: Fast and Efficient Algorithm for Protein Sequence Embedding

Performance comparisons of different settings for GDDA-BLAST and Adaptive GDDA-BLAST. (a) Fold recognition performance of GDDA-BLAST when using different coverage threshold (with pairwise identity 10% threshold). It shows that %coverage is helpful to filter out noisy embedded alignments (no coverage threshold vs. 60% coverage thresholds). (b) Fold recognition performance of Adaptive GDDA-BLAST when using different distance metrics. Mutual information is estimated as described in [32]. 534 sequences of 61 SCOP fold groups from SABmark Twilight zone bechmark set. To calculate the sensitivity at different false positive rates, top-k sequences with the highest similarity to each 534 queries are considered as increasing k from 1.
(0.41 MB EPS)