Skip to main content

Table 2 Performances of the template-based 4-class solvent accessibility predictor (PaleAle_H) compared with a baseline predictor which copies the solvent accessibility of the best template (baseline) and with the ab initio solvent accessibility predictor (PaleAle).

From: Accurate prediction of protein secondary structure and solvent accessibility by consensus combiners of sequence and structure information

maxID baseline PaleAle_H PaleAle
95% 61.1% 64.8% 53.8%
30% 49.2% 57.1% 54.3%
20% 45.6% 55.5% 54.4%
  1. Templates up to 95%, 30% and 20% maximum similarity allowed (maxID) for baseline and PaleAle_H. Performances measured only on residues for which a template has been identified. Results in 5-fold cross-validation on the S2171 set (see text for details).