Mon Dec 27 11:16:48 PST 2010 Kevin Karplus

r_rot_psi_omega_16_mod had the highest mutual information with amino
acids in my tests in ../../compare-real/comparisons/ so I'll try using
it as a design alphabet.

First I made sure that ~/pcem/Makefile.models97 had the alphabet in 
the definition of BASIC_SETUP_TARGETS:		
		info/${CHAIN}.stride-mixed.r_rot_psi_omega_16_mod \
Then I did a "serial-make rm-phi-psi basic-setup" with the
dunbrack-30-pc-1763.ids as input. When that was done,
in ../training-data I ran

make dunbrack-30pc-1763-1.r_rot_psi_omega_16_mod-aa dunbrack-30pc-1763-2.r_rot_psi_omega_16_mod-aa dunbrack-30pc-1763-3.r_rot_psi_omega_16_mod-aa

Then I ran (in ./)
	make dirs
to create the directories, and created
	networks/Gr_rot_psi_omega_16_mod-5-20-5-20-5-aa-empty.net 
based on the similar network in ../pb_rev/networks/


Then I ran 
	make trAB-mult50

Tue Dec 28 07:10:24 PST 2010 Kevin Karplus

bit gain (from quality-reports/)			
epochs	arch			tr12avg	tr12best	tr23avg	tr23best	tr31avg	tr31best
50	bys-3-11-5-11-7-11-9	0.21049	0.2630		0.20551	0.2719		0.20105	0.2640
150	bys-3-11-5-11-7-11-9	0.33736	0.3509		0.33729	0.3522		0.33606	0.3437
250	bys-3-11-5-11-7-11-9	0.37357	0.3759		0.3717	0.3762		0.3639	0.3605
test	bys-3-11-5-11-7-11-9		0.3647			0.3663			0.3665

r_rot_psi_omega_16_mod:
50	5-20-5-20-5		0.24273	0.2896		0.24754	0.2914		0.23643	0.2926


Note: at 50 epochs the r_rot_psi_omega_16_mod alphabet is clearly
beating the bys alphabet.

Wed Dec 29 23:47:16 PST 2010 Kevin Karplus

r_rot_psi_omega_16_mod:
50	5-20-5-20-5		0.24273	0.2896		0.24754	0.2914		0.23643	0.2926
150	5-20-5-20-5		0.35659	0.3650		0.35853	0.3708		0.35018	0.3602

This alphabet is still looking promising after the second round of training.

Thu Dec 30 09:26:22 PST 2010 Kevin Karplus
r_rot_psi_omega_16_mod:
50	5-20-5-20-5		0.24273	0.2896		0.24754	0.2914		0.23643	0.2926
150	5-20-5-20-5		0.35659	0.3650		0.35853	0.3708		0.35018	0.3602
250	5-20-5-20-5		0.39713	0.3980		0.39683	0.3964		0.38857	0.3889
test	5-20-5-20-5			0.3883			0.3896			0.3876

sequence recovery
test	bys-5-20-5-20-5			19.33%			19.23%			19.39%
test	bys-3-11-5-11-7-11-9		19.28%			19.41%			19.58%
test	5-20-5-20-5			19.82%			19.76%			20.02%

The average sequence recovery of 19.87% is better than the best
previous (19.423% for bys-3-11-5-11-7-11-9).

Fri Dec 31 08:09:18 PST 2010 Kevin Karplus

Did
make -k trAB-train-rdb.scripts trAB-train-rdb.run
to make RDB files for testing ways of evaluating design methods.