. 2016 Dec 31;19(3):482–494. doi: 10.1093/bib/bbw129

Table 1.

Method comparison based on Q3 using newly released structures (TS115) and CASP12 targets (15 proteins) for secondary structure prediction

Data set Method	TS115		CASP12		Server location
Data set Method	Q3	P-value^a	Q3	P-value^a	Server location
Jpred4	0.771^b	0.0007	0.751	0.04	http://www.compbio.dundee.ac.uk/jpred4/index.html
SPINE X	0.801	0.0002	0.769	0.006	http://sparks-lab.org/SPINE-X/
PSIPRED 3.3	0.802	0.12	0.780	0.19	http://distillf.ucd.ie/porterpaleale/
SCORPION	0.817	0.45	0.805	0.44	Stand-alone version from http://hpcr.cs.odu.edu/c3scorpion/
SPIDER2	0.819	NA	0.798	NA	http://sparks-lab.org/server/SPIDER2/
PORTER 4.0	0.820	0.17	0.798	0.67	http://distillf.ucd.ie/porterpaleale/
DeepCNF	0.823	0.01	0.821	0.14	http://raptorx2.uchicago.edu/StructurePropertyPred/predict/

Note.

^aPaired t-test from SPIDER2.

^bJpred only predicts the sequence <800 residues. For TS115, there is one sequence (5hdtA) with 1085 residues. 5hdtA was divided into two chains with 800 residues and 285 residues, respectively.