. Author manuscript; available in PMC: 2013 Jan 30.

Published in final edited form as: J Comput Chem. 2011 Nov 2;33(3):259–267. doi: 10.1002/jcc.21968

Table 8.

The mean absolute error (MAE) and the mean error (ME) between predicted and actual secondary structure contents for individual proteins

Dataset	Error Type	MAE			ME

		%H	%E	%C	%H	%E	%C
1833	PSIPRED	5.8±5.2	4.3±4.9	8.0±6.2	−4.2±6.6	−2.0±6.2	6.2±8.0
	This work	4.5±5.3	4.3±5.0	5.8±5.9	−0.6±6.8	−2.2±6.2	2.7±7.8
CASP9	PSIPRED	5.1±4.3	4.5±4.3	8.0±5.4	−3.7±5.5	−3.0±5.4	6.6±7.0
	This work	4.9±4.2	4.7±4.5	5.5±4.8	0.5±6.4	−3.1±5.7	2.6±6.8

Error bars give the standard deviations from the averaged ME and MAE.