PFRMAT SS 
TARGET T0074 
AUTHOR 6140-7890-6093 
METHOD Chandonia & Karplus, PROTEINS 1998, submitted. 
METHOD Predicted from 28 sequences (finally we get a target  
METHOD with good multi-sequence info!) using a jury of neural 
METHOD networks. 
METHOD Expected accuracy is 82.3% for this sequence. 
METHOD Unlike other methods, our NN's don't put any beta strands 
METHOD into the sequence; if we knew there was some, it 
METHOD would go around residues 25-26 (FL).  For these 2 
METHOD residues, the estimated chance of coil is about 5% 
METHOD greater, so coil was predicted.  It would be nice 
METHOD if the CASP submission format allowed for predicting all 3  
METHOD probabilities (H, E, C) at each position instead of just 
METHOD estimating the probability of the prediction being correct. 
MODEL 1 
P C 0.96 
W C 0.82 
A C 0.70 
V C 0.59 
K C 0.64 
P C 0.51 
E H 0.67 
D H 0.72 
K H 0.79 
A H 0.83 
K H 0.92 
Y H 0.92 
D H 0.93 
A H 0.93 
I H 0.91 
F H 0.89 
D H 0.81 
S H 0.61 
L C 0.63 
S C 0.81 
P C 0.87 
V C 0.91 
N C 0.90 
G C 0.78 
F C 0.52 
L C 0.46 
S C 0.53 
G H 0.46 
D H 0.60 
K H 0.63 
V H 0.65 
K H 0.74 
P H 0.74 
V H 0.78 
L H 0.80 
L H 0.76 
N H 0.58 
S C 0.70 
K C 0.90 
L C 0.92 
P C 0.75 
V H 0.78 
D H 0.91 
I H 0.92 
L H 0.90 
G H 0.89 
R H 0.85 
V H 0.76 
W H 0.66 
E H 0.56 
L H 0.44 
S C 0.61 
D C 0.79 
I C 0.84 
D C 0.86 
H C 0.88 
D C 0.90 
G C 0.88 
M C 0.82 
L C 0.77 
D C 0.71 
R H 0.81 
D H 0.96 
E H 0.98 
F H 0.98 
A H 0.98 
V H 0.98 
A H 0.98 
M H 0.98 
F H 0.98 
L H 0.98 
V H 0.98 
Y H 0.96 
C H 0.93 
A H 0.83 
L H 0.61 
E C 0.68 
K C 0.91 
E C 0.95 
P C 0.95 
V C 0.94 
P C 0.94 
M C 0.92 
S C 0.92 
L C 0.94 
P C 0.95 
P C 0.93 
A C 0.92 
L C 0.94 
V C 0.95 
P C 0.96 
P C 0.96 
S C 0.97 
K C 0.94 
R C 0.92 
K C 0.93 
T C 0.94 
W C 0.98 
END 
