Tue May 20 20:08:02 PDT 2008 T0409 Make started Tue May 20 20:08:32 PDT 2008 Running on cheep.cse.ucsc.edu Tue May 20 21:06:42 PDT 2008 Kevin Karplus There do not seem to be any templates getting low e-values. This may be a template-free modeling problem. Tue May 20 22:11:18 PDT 2008 Kevin Karplus On the other hand, 3 of the top 4 hits are for b.40.4.5, so there may be a distant template after all. Fri May 23 04:35:32 PDT 2008 Kevin Karplus The b.40.4.5 models look good, but the first 30 residues are not set from the templates. I think that the helix can be packed in better than in the try1-opt3 model. Wed May 28 11:15:02 PDT 2008 SAM-T08-MQAO hand QA T0409 Submitted Wed May 28 11:15:02 PDT 2008 SAM-T08-MQAU hand QA T0409 Submitted Wed May 28 11:15:02 PDT 2008 SAM-T08-MQAC hand QA T0409 Submitted Mon Jun 2 14:46:11 PDT 2008 Kevin Karplus Interestingly, the MQAU method prefers the Zhang-Server models, but the MQAC (consensus) method prefers METATASSER and BAKER-ROBETTA. I should run the metaserver optimizations, with just the try1 costfcn. Mon Jun 2 14:52:26 PDT 2008 Kevin Karplus (metaserver runs started on moai cluster) try1-opt3 has 2 bad breaks, so I should probably do a polishing run with breaks and clashes turned up: T0409.try1-opt3.pdb.gz breaks before (T0409)E54 with cost 3.80461 T0409.try1-opt3.pdb.gz breaks before (T0409)G77 with cost 1.03031 Mon Jun 2 15:01:04 PDT 2008 Kevin Karplus polishing run started as try2 on moai cluster. Mon Jun 2 17:06:28 PDT 2008 Kevin Karplus Both metaserver runs favored SAM-T08-server_TS1 try2 costfcn favores try2-opt3, try1-opt3.gromacs0, MQAC1-opt3, MQAU1-opt3 Mon Jun 2 17:37:33 PDT 2008 Kevin Karplus All the models agree fairly well on D31 on, and the two main lines of prediction give two different solutions for the first 30 residues. METATASSER_TS4 and Zhang-Server_TS2 give two other possible solutions for 1-30. Looking at just predicted secondary structure components of the cost, it looks like good models are try1-opt3.gromacs0 (for pred_alpha_back), try1-opt3 (for ehl2_constraints), try1-opt2 (for pred_n_notor_back and pred_bys_back), and MQAC1_opt1 (for pred_pb_back). Fri Jun 6 03:11:23 PDT 2008 Kevin Karplus Let me do a polishing run that includes METATASSER and Zhang-Server models, as well as the current models, and hope for some productive crossover. For try3, I'm doing a polishing run starting from our models, the metaserver models, and METATASSER, Zhang-Server, and BAKER-ROBETTA models. Fri Jun 6 04:03:40 PDT 2008 Kevin Karplus The try3 run tweaked up try2-opt3, but the results are not more convincing to rosetta. Fri Jun 6 04:15:38 PDT 2008 Kevin Karplus I've decided I don't care much for the METATASSER_TS4 model, but I'll do one more run to see if polishing the Zhang-Server models yields anything of interest. Fri Jun 6 04:32:18 PDT 2008 Kevin Karplus I first tried scoring all the servers with the try4 costfcn, but that doesn't have any missing_atoms term, so liked best a scwrled alignment model (SAM-T06-server_TS2-scwrl) that did not have the N-terminal part. The Zhang-Server models did not score well with try4, because of clashes and breaks, so the try4 polishing run that uses only the Zhang-Server models will be interesting. Fri Jun 6 05:19:52 PDT 2008 Kevin Karplus Nope, try4 is a mess, wih the barrel torn apart. I don't think this is worth fussing with more. I'll submit ReadConformPDB T0409.try3-opt3.pdb ReadConformPDB T0409.MQAC1-opt3.pdb