# This file is the result of combining several RDB files, specifically
# miniAGRP.t2k.dssp-ebghstl.rdb  (weight 1.53986)
# miniAGRP.t2k.stride-ebghtl.rdb  (weight 1.24869)
# miniAGRP.t2k.str2.rdb  (weight 1.54758)
# miniAGRP.t2k.alpha.rdb  (weight 0.659012)
# These files were combined by translating their predictions into EHL
# predictions with tables generated by compare-real, and then combining
# those predictions with weights proportional to their mutual information
# with the EHL alphabet.  The comments from the individual files follow.
#
# Comments from miniAGRP.t2k.dssp-ebghstl.rdb
# ============================================
# TARGET miniAGRP
# Using neural net t2k-5740-IDaaHr-5-15-7-15-9-15-13-ebghstl-seeded.net
# This is a 4-layer network, with 
#	window	units
#	    5	   15
#	    7	   15
#	    9	   15
#	   13	    7 (1 EBGHSTL )
# The input amino acid frequencies were determined from
# alignment miniAGRP.t2k-thin90.a2m.gz
# with weighted counts, using HenikoffWeight(-1 bits/column, 1)
# Counts were regularized to probabilities using
# /projects/compbio/lib/recode3.20comp
# Total sequence weight for alignment was 9
#
# ============================================
# Comments from miniAGRP.t2k.stride-ebghtl.rdb
# ============================================
# TARGET miniAGRP
# Using neural net t2k-5651-IDaaHr-5-15-7-15-9-15-13-ebghtl-seeded.net
# This is a 4-layer network, with 
#	window	units
#	    5	   15
#	    7	   15
#	    9	   15
#	   13	    6 (1 EBGHTL )
# The input amino acid frequencies were determined from
# alignment miniAGRP.t2k-thin90.a2m.gz
# with weighted counts, using HenikoffWeight(-1 bits/column, 1)
# Counts were regularized to probabilities using
# /projects/compbio/lib/recode3.20comp
# Total sequence weight for alignment was 9
#
# ============================================
# Comments from miniAGRP.t2k.str2.rdb
# ============================================
# TARGET miniAGRP
# Using neural net t2k-5651-IDaaHr-5-15-7-15-9-15-13-str2-from-empty.net
# This is a 4-layer network, with 
#	window	units
#	    5	   15
#	    7	   15
#	    9	   15
#	   13	   13 (1 str2 )
# The input amino acid frequencies were determined from
# alignment miniAGRP.t2k-thin90.a2m.gz
# with weighted counts, using HenikoffWeight(-1 bits/column, 1)
# Counts were regularized to probabilities using
# /projects/compbio/lib/recode3.20comp
# Total sequence weight for alignment was 9
#
# ============================================
# Comments from miniAGRP.t2k.alpha.rdb
# ============================================
# TARGET miniAGRP
# Using neural net t2k-5651-IDaaHr-5-15-7-15-9-15-13-alpha-seeded.net
# This is a 4-layer network, with 
#	window	units
#	    5	   15
#	    7	   15
#	    9	   15
#	   13	   11 (1 ABCDEFGHIST )
# The input amino acid frequencies were determined from
# alignment miniAGRP.t2k-thin90.a2m.gz
# with weighted counts, using HenikoffWeight(-1 bits/column, 1)
# Counts were regularized to probabilities using
# /projects/compbio/lib/recode3.20comp
# Total sequence weight for alignment was 9
#
# ============================================
Pos	AA	E	H	C
10N	1S	5N	5N	5N
1	C	0.1426	0.0572	0.8001
2	V	0.2294	0.0787	0.6919
3	R	0.2438	0.0978	0.6584
4	L	0.1855	0.2177	0.5968
5	H	0.1627	0.2753	0.5621
6	E	0.1333	0.3141	0.5526
7	S	0.1455	0.2551	0.5995
8	C	0.1381	0.1882	0.6737
9	L	0.1265	0.1181	0.7555
10	G	0.0907	0.0804	0.8289
11	Q	0.0861	0.0699	0.8440
12	Q	0.1635	0.0518	0.7848
13	V	0.1413	0.0362	0.8225
14	P	0.1562	0.0543	0.7894
15	C	0.2604	0.0647	0.6749
16	C	0.2578	0.0488	0.6934
17	D	0.1504	0.0605	0.7892
18	P	0.0488	0.3188	0.6324
19	A	0.0677	0.3294	0.6028
20	A	0.0882	0.3027	0.6091
21	T	0.1883	0.2693	0.5423
22	C	0.2711	0.3543	0.3746
23	Y	0.3407	0.4000	0.2593
24	C	0.3779	0.4221	0.2000
25	R	0.4160	0.4199	0.1641
26	F	0.3837	0.4142	0.2021
27	F	0.3384	0.3909	0.2707
28	N	0.3082	0.3254	0.3664
29	A	0.3449	0.2916	0.3635
30	F	0.4845	0.2094	0.3060
31	C	0.4404	0.1835	0.3761
32	Y	0.4410	0.1115	0.4474
33	C	0.2454	0.0787	0.6759
34	R	0.0422	0.0496	0.9082