Thu Jun 15 09:11:04 PDT 2006 T0333 Make started Thu Jun 15 09:12:06 PDT 2006 Running on shaw.cse.ucsc.edu Thu Jun 15 09:18:56 PDT 2006 Kevin Karplus No strong hits in PDB with BLAST. Weak hits to 1zlyA, 1meoA, and 1mejA (31% over 48 residues, E-value 0.84). Thu Jun 15 17:26:16 PDT 2006 Kevin Karplus Strong HMM hits to 2c1xA (2.5e-26), 2acvA (7e-26), 1f0kA(4e-25), ... and other c.87.1.* domains. Thu Jun 15 22:21:01 PDT 2006 Kevin Karplus Try1-opt2 does not look bad. It is very close to the top 5 templates, though different enough to need some refinement. There are a lot of really bad breaks that still need to be closed. I'm going to do another run from the alignments, with a higher break penalty and with sheets and helices from try1-opt2. Fri Jun 16 07:29:21 PDT 2006 Kevin Karplus try2-opt2 does not score quite as well as try1-opt2 did with the try2 costfcn. It also does not seem to superimpose as well with the top few alignments. The problem is that the constraints for holding the sheets were not strong enough to overpower the break costs. Perhaps I should get another starting point with yet another variant of the cost function, with the break costs intermediate between try1 and try2. (started as try3 on the farm cluster) Fri Jun 30 13:42:43 PDT 2006 Kevin Karplus The try3-opt2 result is better with the try3 costfcn and the try2 costfcn, though the try2 costfcn prefers try3-opt2.repack-nonPC, because of the reduced clashes. Rosetta hates repacking all three (try1-opt2, try2-opt2, and try3-opt2), but complains least about try2. Fri Jun 30 13:52:28 PDT 2006 Kevin Karplus try3-opt2 looks like an ok start, but there are still breaks to close, including some very bad ones: T0333.try3-opt2.pdb.gz breaks before (T0333)P211 with cost 12.316 T0333.try3-opt2.pdb.gz breaks before (T0333)Q311 with cost 6.40554 T0333.try3-opt2.pdb.gz breaks before (T0333)R264 with cost 6.39258 T0333.try3-opt2.pdb.gz breaks before (T0333)D28 with cost 5.75686 T0333.try3-opt2.pdb.gz breaks before (T0333)D333 with cost 4.54768 T0333.try3-opt2.pdb.gz breaks before (T0333)G342 with cost 4.18776 T0333.try3-opt2.pdb.gz breaks before (T0333)A37 with cost 3.84173 T0333.try3-opt2.pdb.gz breaks before (T0333)V332 with cost 3.83811 I'll start a polishing run from the models, with break penalties turned up high. Fri Jun 30 13:58:47 PDT 2006 Kevin Karplus try4 started on whidbey. Fri Jun 30 14:16:59 PDT 2006 Firas Khatib I will start working on this protein since it's soft deadline is july 6th, in less than a week. Fri Jun 30 15:50:18 PDT 2006 Kevin Karplus try4 restarted, since the read-pdb.under file was missing and the first run failed. Fri Jun 30 15:54:44 PDT 2006 Firas Khatib try3-opt2-repack also scores best with the unconstrained cost function Mon Jul 3 13:24:02 PDT 2006 Kevin Karplus Correction: it is try4-opt2-repack that scored best with try4 and try4-opt2 that scores best with unconstrained costfcn. There are still some awful breaks: Conformation[17] T0333.try4-opt2.pdb.gz has 45 breaks T0333.try4-opt2.pdb.gz breaks before (T0333)P211 with cost 12.0532 T0333.try4-opt2.pdb.gz breaks before (T0333)Q311 with cost 5.86419 T0333.try4-opt2.pdb.gz breaks before (T0333)R264 with cost 4.56848 T0333.try4-opt2.pdb.gz breaks before (T0333)L82 with cost 3.48264 T0333.try4-opt2.pdb.gz breaks before (T0333)G196 with cost 3.46485 T0333.try4-opt2.pdb.gz breaks before (T0333)D333 with cost 3.1009 T0333.try4-opt2.pdb.gz breaks before (T0333)V332 with cost 2.53328 T0333.try4-opt2.pdb.gz breaks before (T0333)D28 with cost 1.48856 T0333.try4-opt2.pdb.gz breaks before (T0333)T348 with cost 1.44274 T0333.try4-opt2.pdb.gz breaks before (T0333)A37 with cost 0.673429 Perhaps I should try the trick of reoptimizing from the gromacs models. Mon Jul 3 13:41:13 PDT 2006 Kevin Karplus try5 started on farm cluster. Tue Jul 4 15:39:30 PDT 2006 Kevin Karplus Because gromacs doesn't run on the cluster, I redid the final steps of try5. Best-scoring now try5-opt2 Rosetta likes best decoys/T0333.try5-opt2.gromacs0.repack-nonPC.pdb.gz There are still some bad breaks: Conformation[23] T0333.try5-opt2.pdb.gz has 36 breaks T0333.try5-opt2.pdb.gz breaks before (T0333)D252 with cost 2.0948 T0333.try5-opt2.pdb.gz breaks before (T0333)H160 with cost 0.844147 T0333.try5-opt2.pdb.gz breaks before (T0333)D106 with cost 0.776801 T0333.try5-opt2.pdb.gz breaks before (T0333)K159 with cost 0.695494 Conformation[22] T0333.try5-opt2.gromacs0.repack-nonPC.pdb.gz has 22 breaks T0333.try5-opt2.gromacs0.repack-nonPC.pdb.gz breaks before (T0333)D252 with cost 1.84521 T0333.try5-opt2.gromacs0.repack-nonPC.pdb.gz breaks before (T0333)Q62 with cost 1.01437 T0333.try5-opt2.gromacs0.repack-nonPC.pdb.gz breaks before (T0333)D106 with cost 0.825615 T0333.try5-opt2.gromacs0.repack-nonPC.pdb.gz breaks before (T0333)K159 with cost 0.790926 T0333.try5-opt2.gromacs0.repack-nonPC.pdb.gz breaks before (T0333)G226 with cost 0.641153 T0333.try5-opt2.gromacs0.repack-nonPC.pdb.gz breaks before (T0333)R307 with cost 0.615284 Tue Jul 4 15:50:24 PDT 2006 Kevin Karplus try5-opt2 is based on try2-opt2.gromacs0, which I don't like as well as the try3 and try4 models. I'm going to redo the polishing, but using only try3 and try4 starting points. Tue Jul 4 16:07:21 PDT 2006 Kevin Karplus try6 running on lopez. Tue Jul 4 23:53:43 PDT 2006 Kevin Karplus Submitted with comment Model 1 is try6-opt2. Model 2 is try5-opt2, a somewhat different alignment. Model 3 is try1-opt2, the first, fully automatic, model. Model 4 is sidechain replacment by SCWRL on an alignment to 2c1xA. Model 5 is sidechain replacment by SCWRL on an alignment to 2acvA. Fri Jul 7 23:48:22 PDT 2006 Kevin Karplus I see no new comments here, so I assume that Firas is willing to stick with the preliminary submission.