Mon Aug 12 11:29:34 PDT 2002 t0178 Obvious homology to 1jcjA, 1jclA (c.1.10.1) 20 August 2002 Oscar Hur & Jenny Draper Ran try2 with upped values for reduceBreak and closeGap, to try to fix the massive number of breaks. It's a little better, but still needs work. Date: Tue, 20 Aug 2002 15:34:19 -0700 (PDT) From: Oscar Hur To: Kevin Karplus Subject: t0178 Dear Dr Karplus Jenny has helped me looking at T0178 aldolase. There are some breaks in the structure for the alpha-beta motifs. I want to know if we can add some constraints to the structure. We have already done try2 yesterday. Do you want the structure to be a specific conformation? Oscar Date: Tue, 20 Aug 2002 16:41:52 -0700 From: Kevin Karplus To: oscarhur@soe.ucsc.edu Cc: karplus@soe.ucsc.edu, learithe@soe.ucsc.edu Subject: Re: t0178 I just talked with Jenny about your work on t0178. It looks like a fairly easy homology model that is getting the right alignment, so you are now in what I call the "polishing" phase of the process---trying to close up small breaks and jiggle the backbones and sidechains to get tighter packing. I gave some advice to Jenny on this target. When we meet on Thursday we can discuss it further. ------------------------------------------------------------ Wed Aug 21 10:22:09 PDT 2002 Ran VAST yesterday on the try2-opt-scwrl model, and it got a 1.0 Ang rmsd over 208 residues with 1jclA. Interestingly, the VAST alignment moved several of the breaks out of the helices into the loops, so looks like a better starting alignment. Starting try4 from JUST that alignment gets initial alignments that are a little worse than the initial alignments of try1, and substantially worse than the initial conformations of try2 and try3. Still, if the gaps can be healed, they are in better places than in the other conformations, so optimization should be able to get to a better result. 21 Aug Oscar Hur Try5 has overwritten try4, as I forgot to replace "try4" with "try5" in undertaker.script. Try5 should have read the try4 structure though... 22 Aug Oscar Hur Look at the try4 (which is supposed to be try5). The gap, which orginally presents in the previous ones (before try4) between 123 and 124, is still intact. However, there is a gap between 120 and 121 nearby. Try4 is not good. So do try5 using the original alignment and change the parameter of OneRotamer, and remove SS bond values. -------------------------------------------------- 25 Aug 2002 Kevin Karplus Currently the best score is for try4-opt, second best try3-opt. All the try*-opt look like good TIM barrels, and the current scoring function uses no constraints, so we're mainly looking at breaks and hydrophobic burial now. I'll turn pred_alpha2 down, and add SSBOND C38-C57 to the score function, and turn breaks and dry scores up. I'll also move all the T0178.try4 files in . to decoys/oops-T0178.try4... so that they can be included in score-decoys.rdb try3-opt-scwrl is now the best, with try4-opt-scwrl close behind, and the oops-T0178.try4-opt-scwrl just behind that. Let's do an optimization run from all three of them. Since Oscar's try5 run died due to a bug in undertaker, since fixed, let's rerun as try5. From karplus@bray.cse.ucsc.edu Sun Aug 25 13:43:12 2002 Date: Sun, 25 Aug 2002 13:43:10 -0700 From: Kevin Karplus To: oscarhur@soe.ucsc.edu CC: karplus@soe.ucsc.edu Subject: T0178 I've taken T0178 over again, so that I can try to get it out today. Please start on some later protein (perhaps one of the high 180s or low 190s). ------------------------------------------------------------ 27 Aug 2002 02:01 Kevin Karplus I didn't get a chance to look at t0178 earlier today, though it finished the try5 run at 06:23. As expected, try5-opt is the new top score. It's not perfectly packed, but we're no likely to get much better, since there are charged residues in the center of the barrel, and they're going to insist on some solvent. Let's submit it.