
PFRMAT SS
TARGET T0106
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
A C 0.96
A C 0.85
C C 0.70
E C 0.30
P E 0.18
V E 0.20
R E 0.06
I E 0.42
P C 0.78
L C 0.64
C C 0.45
K C 0.64
S C 0.75
L C 0.92
P C 0.90
W C 0.58
E C 0.23
M C 0.33
T C 0.61
K C 0.78
M C 0.72
P C 0.37
N C 0.52
H C 0.24
L C 0.39
H C 0.84
H C 0.88
S C 0.86
T C 0.94
Q H 0.85
A H 0.94
N H 0.95
A H 0.96
I H 0.97
L H 0.95
A H 0.96
M H 0.95
E H 0.93
Q H 0.82
F H 0.63
E H 0.80
G H 0.68
L H 0.77
L H 0.70
G C 0.07
T C 0.71
H C 0.92
C C 0.92
S C 0.86
P H 0.77
D H 0.87
L H 0.90
L H 0.97
F H 0.96
F H 0.96
L H 0.96
C H 0.93
A H 0.91
M H 0.80
Y C 0.16
A C 0.45
P C 0.57
I C 0.55
C C 0.70
T C 0.89
I C 0.82
D C 0.84
F C 0.86
Q C 0.88
H C 0.84
E C 0.80
P C 0.49
I C 0.75
K C 0.75
P C 0.81
C C 0.51
K H 0.67
S H 0.86
V H 0.83
C H 0.71
E H 0.95
R H 0.91
A H 0.83
R H 0.88
Q H 0.77
G H 0.57
C H 0.75
E H 0.81
P H 0.90
I H 0.87
L H 0.92
I H 0.85
K H 0.67
Y C 0.15
R C 0.88
H C 0.88
S C 0.83
W C 0.83
P C 0.87
E C 0.70
S C 0.37
L C 0.32
A C 0.39
C C 0.07
D C 0.02
E C 0.09
L C 0.84
P C 0.80
V C 0.83
Y C 0.70
D C 0.74
R C 0.85
G C 0.85
V C 0.70
C C 0.77
I C 0.73
S C 0.70
P C 0.76
E C 0.78
A C 0.62
I C 0.55
V C 0.70
T C 0.76
A C 0.80
D C 0.89
G C 0.93
A C 0.88
D C 0.96
END

