PFRMAT SS
TARGET T0437
AUTHOR SAM-T06-server 
METHOD This file is the result of combining several RDB files, specifically
METHOD T0437.t04.dssp-ebghstl.rdb  (weight 1.53986)
METHOD T0437.t04.stride-ebghtl.rdb  (weight 1.24869)
METHOD T0437.t04.str2.rdb  (weight 1.54758)
METHOD T0437.t04.alpha.rdb  (weight 0.659012)
METHOD T0437.t2k.dssp-ebghstl.rdb  (weight 1.53986)
METHOD T0437.t2k.stride-ebghtl.rdb  (weight 1.24869)
METHOD T0437.t2k.str2.rdb  (weight 1.54758)
METHOD T0437.t2k.alpha.rdb  (weight 0.659012)
METHOD These files were combined by translating their predictions into EHL
METHOD predictions with tables generated by compare-real, and then combining
METHOD those predictions with weights proportional to their mutual information
METHOD with the EHL alphabet.  The comments from the individual files follow.
METHOD
METHOD Comments from T0437.t04.dssp-ebghstl.rdb
METHOD ============================================
METHOD TARGET T0437
METHOD Using neural net t04-2621-IDGaaH13-3-13-7-13-9-13-11-ebghstl-dssp-seeded.net
METHOD This is a 4-layer network, with 
METHOD	window	units
METHOD	    3	   13
METHOD	    7	   13
METHOD	    9	   13
METHOD	   11	    7 (1 EBGHSTL )
METHOD The input amino acid frequencies were determined from
METHOD alignment T0437.t04-thin90.a2m.gz
METHOD with weighted counts, using HenikoffWeight(1.3 bits/column, 1)
METHOD The weighting was determined by the posterior distribution
METHOD after regularizing with /projects/compbio/lib/recode3.20comp.
METHOD Counts were regularized to probabilities using
METHOD /projects/compbio/lib/recode3.20comp
METHOD Total sequence weight for alignment was 16.5273
METHOD
METHOD ============================================
METHOD Comments from T0437.t04.stride-ebghtl.rdb
METHOD ============================================
METHOD TARGET T0437
METHOD Using neural net t04-2621-IDGaaH13-3-13-7-13-9-13-11-ebghtl-stride-seeded.net
METHOD This is a 4-layer network, with 
METHOD	window	units
METHOD	    3	   13
METHOD	    7	   13
METHOD	    9	   13
METHOD	   11	    6 (1 EBGHTL )
METHOD The input amino acid frequencies were determined from
METHOD alignment T0437.t04-thin90.a2m.gz
METHOD with weighted counts, using HenikoffWeight(1.3 bits/column, 1)
METHOD The weighting was determined by the posterior distribution
METHOD after regularizing with /projects/compbio/lib/recode3.20comp.
METHOD Counts were regularized to probabilities using
METHOD /projects/compbio/lib/recode3.20comp
METHOD Total sequence weight for alignment was 16.5273
METHOD
METHOD ============================================
METHOD Comments from T0437.t04.str2.rdb
METHOD ============================================
METHOD TARGET T0437
METHOD Using neural net t04-2621-IDGaaH13-3-13-7-13-9-13-11-str2-seeded.net
METHOD This is a 4-layer network, with 
METHOD	window	units
METHOD	    3	   13
METHOD	    7	   13
METHOD	    9	   13
METHOD	   11	   13 (1 str2 )
METHOD The input amino acid frequencies were determined from
METHOD alignment T0437.t04-thin90.a2m.gz
METHOD with weighted counts, using HenikoffWeight(1.3 bits/column, 1)
METHOD The weighting was determined by the posterior distribution
METHOD after regularizing with /projects/compbio/lib/recode3.20comp.
METHOD Counts were regularized to probabilities using
METHOD /projects/compbio/lib/recode3.20comp
METHOD Total sequence weight for alignment was 16.5273
METHOD
METHOD ============================================
METHOD Comments from T0437.t04.alpha.rdb
METHOD ============================================
METHOD TARGET T0437
METHOD Using neural net t04-2621-IDGaaH13-3-13-7-13-9-13-11-alpha-seeded.net
METHOD This is a 4-layer network, with 
METHOD	window	units
METHOD	    3	   13
METHOD	    7	   13
METHOD	    9	   13
METHOD	   11	   11 (1 ABCDEFGHIST )
METHOD The input amino acid frequencies were determined from
METHOD alignment T0437.t04-thin90.a2m.gz
METHOD with weighted counts, using HenikoffWeight(1.3 bits/column, 1)
METHOD The weighting was determined by the posterior distribution
METHOD after regularizing with /projects/compbio/lib/recode3.20comp.
METHOD Counts were regularized to probabilities using
METHOD /projects/compbio/lib/recode3.20comp
METHOD Total sequence weight for alignment was 16.5273
METHOD
METHOD ============================================
METHOD Comments from T0437.t2k.dssp-ebghstl.rdb
METHOD ============================================
METHOD TARGET T0437
METHOD Using neural net t2k-5740-IDaaHr-5-15-7-15-9-15-13-ebghstl-seeded.net
METHOD This is a 4-layer network, with 
METHOD	window	units
METHOD	    5	   15
METHOD	    7	   15
METHOD	    9	   15
METHOD	   13	    7 (1 EBGHSTL )
METHOD The input amino acid frequencies were determined from
METHOD alignment T0437.t2k-thin90.a2m.gz
METHOD with weighted counts, using HenikoffWeight(-1 bits/column, 1)
METHOD Counts were regularized to probabilities using
METHOD /projects/compbio/lib/recode3.20comp
METHOD Total sequence weight for alignment was 52
METHOD
METHOD ============================================
METHOD Comments from T0437.t2k.stride-ebghtl.rdb
METHOD ============================================
METHOD TARGET T0437
METHOD Using neural net t2k-5651-IDaaHr-5-15-7-15-9-15-13-ebghtl-seeded.net
METHOD This is a 4-layer network, with 
METHOD	window	units
METHOD	    5	   15
METHOD	    7	   15
METHOD	    9	   15
METHOD	   13	    6 (1 EBGHTL )
METHOD The input amino acid frequencies were determined from
METHOD alignment T0437.t2k-thin90.a2m.gz
METHOD with weighted counts, using HenikoffWeight(-1 bits/column, 1)
METHOD Counts were regularized to probabilities using
METHOD /projects/compbio/lib/recode3.20comp
METHOD Total sequence weight for alignment was 52
METHOD
METHOD ============================================
METHOD Comments from T0437.t2k.str2.rdb
METHOD ============================================
METHOD TARGET T0437
METHOD Using neural net t2k-5651-IDaaHr-5-15-7-15-9-15-13-str2-from-empty.net
METHOD This is a 4-layer network, with 
METHOD	window	units
METHOD	    5	   15
METHOD	    7	   15
METHOD	    9	   15
METHOD	   13	   13 (1 str2 )
METHOD The input amino acid frequencies were determined from
METHOD alignment T0437.t2k-thin90.a2m.gz
METHOD with weighted counts, using HenikoffWeight(-1 bits/column, 1)
METHOD Counts were regularized to probabilities using
METHOD /projects/compbio/lib/recode3.20comp
METHOD Total sequence weight for alignment was 52
METHOD
METHOD ============================================
METHOD Comments from T0437.t2k.alpha.rdb
METHOD ============================================
METHOD TARGET T0437
METHOD Using neural net t2k-5651-IDaaHr-5-15-7-15-9-15-13-alpha-seeded.net
METHOD This is a 4-layer network, with 
METHOD	window	units
METHOD	    5	   15
METHOD	    7	   15
METHOD	    9	   15
METHOD	   13	   11 (1 ABCDEFGHIST )
METHOD The input amino acid frequencies were determined from
METHOD alignment T0437.t2k-thin90.a2m.gz
METHOD with weighted counts, using HenikoffWeight(-1 bits/column, 1)
METHOD Counts were regularized to probabilities using
METHOD /projects/compbio/lib/recode3.20comp
METHOD Total sequence weight for alignment was 52
METHOD
METHOD ============================================
MODEL 1
M C 0.88
K C 0.74
D C 0.68
V C 0.70
V C 0.72
D H 0.53
K H 0.56
C H 0.55
S C 0.45
T C 0.58
K C 0.71
G C 0.75
C C 0.65
A C 0.45
I E 0.53
D E 0.68
I E 0.66
G E 0.69
T E 0.82
V E 0.84
I E 0.80
D E 0.50
N C 0.78
D C 0.83
N C 0.77
C C 0.50
T E 0.84
S E 0.90
K E 0.92
F E 0.91
S E 0.91
R E 0.88
F E 0.82
F E 0.58
A C 0.80
T C 0.89
R H 0.85
E H 0.93
E H 0.95
A H 0.95
E H 0.96
S H 0.95
F H 0.95
M H 0.95
T H 0.95
K H 0.95
L H 0.95
K H 0.96
E H 0.95
L H 0.94
A H 0.92
A H 0.91
A H 0.86
A H 0.64
S C 0.70
S C 0.64
A C 0.69
D C 0.75
E C 0.80
G C 0.71
A C 0.51
S E 0.63
V E 0.73
A E 0.74
Y E 0.73
K E 0.82
I E 0.84
K E 0.75
D E 0.72
L C 0.50
E C 0.69
G C 0.84
Q C 0.63
V E 0.81
E E 0.90
L E 0.90
D E 0.91
A E 0.91
A E 0.92
F E 0.92
T E 0.92
F E 0.86
S E 0.81
C E 0.58
Q E 0.47
A H 0.40
E H 0.58
M H 0.65
I H 0.71
I H 0.76
F H 0.79
E H 0.82
L H 0.80
S H 0.73
L H 0.65
R H 0.61
S C 0.49
L C 0.62
A C 0.86
END