PFRMAT SS 
TARGET T0052 
AUTHOR 1751-3146-3362 
METHOD Using neural net fssp-guide-12345-IDaa13-9-6-11-9-3-8-7-ehl-seeded-trained.net 
METHOD This is a 4-layer network, with amino acid frequencies, insertions, 
METHOD and deletions as inputs, and the following layers: 
METHOD	window	units 
METHOD	    9	    6 
METHOD	   11	    9 
METHOD	    3	    8 
METHOD	    7	    3 (EHC codes) 
METHOD  
METHOD The input amino acid frequencies were determined from weighted counts 
METHOD (1.3 bits/column) and the recode2.20comp Dirichlet mixture regularizer. 
METHOD The input alignment (using the SAM/T98 method) had only one sequence, 
METHOD so we expect this prediction to be quite poor---probably only about 
METHOD Q3=65%.  The neural net used all available training data, and so 
METHOD no independent assesment of its quality is available. 
METHOD  
METHOD It is unclear whether the helix fragments at QTcyNSA  
METHOD should be merged into a single helix or omitted. 
METHOD  
MODEL 1 
L C 0.78 
G C 0.80 
K C 0.64 
F C 0.54 
S C 0.42 
Q H 0.42 
T H 0.41 
C C 0.41 
Y C 0.47 
N H 0.44 
S H 0.55 
A H 0.50 
I C 0.46 
Q C 0.56 
G C 0.63 
S C 0.59 
V E 0.47 
L E 0.50 
T E 0.43 
S C 0.49 
T C 0.57 
C C 0.54 
E C 0.47 
R C 0.54 
T C 0.74 
N C 0.87 
G C 0.91 
G C 0.88 
Y C 0.87 
N C 0.83 
T C 0.79 
S C 0.74 
S C 0.65 
I C 0.60 
D C 0.65 
L C 0.53 
N H 0.54 
S H 0.69 
V H 0.63 
I H 0.50 
E C 0.48 
N C 0.73 
V C 0.85 
D C 0.83 
G C 0.76 
S C 0.67 
L C 0.47 
K E 0.53 
W C 0.46 
Q C 0.69 
P C 0.77 
S C 0.67 
N H 0.49 
F H 0.53 
I H 0.61 
E H 0.63 
T H 0.49 
C C 0.65 
R C 0.70 
N C 0.71 
T C 0.71 
N C 0.62 
L C 0.53 
A C 0.66 
G C 0.80 
S C 0.70 
S H 0.74 
E H 0.77 
L H 0.83 
A H 0.92 
A H 0.94 
E H 0.90 
C H 0.77 
K H 0.68 
T H 0.74 
R H 0.75 
A H 0.72 
Q H 0.69 
Q H 0.57 
F H 0.37 
V E 0.39 
S C 0.45 
T C 0.44 
K C 0.43 
I C 0.55 
N C 0.77 
L C 0.78 
D C 0.64 
D H 0.53 
H H 0.48 
I C 0.48 
A C 0.47 
N C 0.58 
I C 0.65 
D C 0.70 
G C 0.72 
T C 0.71 
L C 0.55 
K C 0.48 
Y C 0.50 
E C 0.59 
END 
