
PFRMAT SS
TARGET T0110
AUTHOR 9521-1365-5893
METHOD Program Name: PSIPRED
METHOD Synopsis: Neural network prediction based on PSIBLAST output
METHOD Five neural networks are used to predict secondary structure directly
METHOD from the profiles generated by PSIBLAST. Networks 1-4 predict
METHOD secondary structure, the 5th network filters the output of the first
METHOD level networks. The first level networks have 17*21 inputs and 75 hidden
METHOD units, the second level network has 17*3 inputs and 55 hidden units.
METHOD The networks have been trained on a set of 2937 proteins with
METHOD early-stopped training. 10% of the training data was used to detect
METHOD convergence.
METHOD The target sequence was scanned against a non-redundant databank of
METHOD filtered protein sequences (> 500,000 sequences), with 3 PSIBLAST
METHOD iterations.
MODEL  1
M C 0.98
A C 0.86
R C 0.65
E C 0.18
F H 0.41
K H 0.58
R H 0.86
S H 0.88
D H 0.94
R H 0.96
V H 0.96
A H 0.96
Q H 0.96
E H 0.96
I H 0.96
Q H 0.96
K H 0.96
E H 0.96
I H 0.96
A H 0.95
V H 0.96
I H 0.94
L H 0.94
Q H 0.84
R H 0.66
E H 0.47
V C 0.30
K C 0.59
D C 0.80
P C 0.67
R C 0.78
I C 0.81
G C 0.74
M E 0.06
V E 0.55
T E 0.84
V E 0.94
S E 0.85
D E 0.94
V E 0.95
E E 0.97
V E 0.88
S E 0.46
S C 0.84
D C 0.91
L C 0.11
S E 0.29
Y E 0.50
A E 0.85
K E 0.92
I E 0.94
F E 0.96
V E 0.95
T E 0.92
F E 0.59
L C 0.05
F C 0.89
D C 0.92
H C 0.66
D C 0.85
E H 0.64
M H 0.87
A H 0.88
I H 0.95
E H 0.96
Q H 0.95
G H 0.97
M H 0.96
K H 0.95
G H 0.94
L H 0.90
E H 0.91
K H 0.74
A H 0.40
S H 0.68
P H 0.58
Y H 0.94
I H 0.95
R H 0.95
S H 0.96
L H 0.95
L H 0.90
G H 0.77
K H 0.23
A C 0.23
M C 0.85
R C 0.66
L E 0.42
R E 0.81
I E 0.93
V E 0.54
P E 0.61
E E 0.74
I E 0.92
R E 0.94
F E 0.96
I E 0.93
Y E 0.71
D E 0.36
Q C 0.61
S C 0.21
L C 0.45
V C 0.76
E C 0.15
G H 0.20
M H 0.74
R H 0.87
M H 0.93
S H 0.95
N H 0.96
L H 0.96
V H 0.96
T H 0.96
N H 0.96
V H 0.94
V H 0.94
R H 0.93
E H 0.83
D H 0.79
E H 0.66
K H 0.50
K H 0.16
H C 0.50
V C 0.76
E C 0.89
E C 0.88
S C 0.87
N C 0.97
END

