Figure 5.

Distribution of 4-class (4%, 25% and 50% exposed thresholds) solvent accessibility prediction accuracy as a function of sequence similarity to the best hit in PSI-BLAST templates. The blue bars represent predictions using templates (maximal sequence similarity allowed is 95%), the red bars template-less predictions. See text for details.

Pollastri et al. BMC Bioinformatics 2007 8:201   doi:10.1186/1471-2105-8-201
Download authors' original image