Skip to main content
. 2012 Jun 9;28(12):i59–i66. doi: 10.1093/bioinformatics/bts213

Table 6.

The contribution of protein features and the impact of different training methods

Training approaches ML MAP Quality-sensitive
Profile + SS3 1536.01 1578.60 1612.44
Profile + SS3 + SS8 1567.81 1595.72 1637.64
Profile + SS3 + SA 1606.68 1616.04 1662.12
Profile + SS3 + SS8 + SA 1633.32 1664.28 1692.17

The results are obtained on our In-House benchmark. SS3 and SS8 indicate the 3-class and 8-class secondary structure, respectively. SA indicates the 3-class solvent accessibility.