Thu May 18 09:57:20 PDT 2006 T0294 Make started Thu May 18 09:58:22 PDT 2006 Running on camano.cse.ucsc.edu Thu May 18 15:36:35 PDT 2006 Kevin Karplus This is a comparative modeling target. Even t2k gets 23 pdb files in the multiple alignment. Probably need to use BLAST to figure out which are the best templates. The HMM scoring likes 1gdhA 2nacA 1j4aA 1dxy 1wwkA 2g76A 1psdA 2d0iA ... but these may be more representative of the consensus than the closest inidividual hits. The t2k best hits (which are a little more closely focussed) start with 1wwkA 2g76A 1psdA 1gdhA 2d0iA 2nacA Make started Thu May 18 20:02:42 PDT 2006 Running on lopez.cse.ucsc.edu I accidentally killed the job on camano, since another job on it was thrashing. I restarted on lopez. The blast hits are excellent for this target: T0294 1gdhA 34.55 301 191 4 8 306 3 299 8.1e-42 166.0 T0294 1wwkA 37.71 236 142 3 63 298 56 286 4.2e-38 153.7 T0294 1ygyA 33.33 285 184 4 44 328 37 315 2.5e-35 144.4 T0294 2g76A 31.94 263 169 5 43 303 59 313 5.8e-32 133.3 T0294 2d0iA 31.17 324 207 8 8 326 4 316 6.5e-31 129.8 T0294 2go1A 29.84 258 177 4 70 325 110 365 1.3e-23 105.5 T0294 2nacA 29.84 258 177 4 70 325 109 364 1.3e-23 105.5 T0294 2gsdA 27.52 258 183 4 70 325 110 365 3.6e-21 97.44 The aborted try1 run seems to have favored 2d0iA as a template, though the other top blast hits were all available in all-align.a2m. Fri May 19 09:07:31 PDT 2006 Kevin Karplus This looks like a 2-domain protein, with the second domain inserted in the middle of the first domain. The model looks pretty good, though all the conserved residues are in the contiguous inserted domain. The rr prediction maybe misleading (or our current prediction may be drastically wrong), as the only rr constraint shown in the rasmol script on the c-terminal piece implies that it is part of the inserted domain, rather than first domain. Fri May 19 12:33:18 PDT 2006 Kevin Karplus 2d0iA was the favored alignment for the completed try1 run also. Thu May 25 15:09:14 PDT 2006 Kevin Karplus Why are C288, I290, and P292 exposed despite high prediction of burial? Look at templates. Is this a dimerization interface? Mon Jun 5 14:42:48 PDT 2006 Kevin Karplus The top scorers with the try1 costfn (including servers) are SAM_T06_server_TS1 SAM_T06_server_TS1-scwrl T0294.try1-opt2.pdb.gz T0294.try1-opt2.repack-n ROBETTA_TS1 SAM_T06_server_TS5-scwrl mGen-3D_TS1-scwrl ROBETTA_TS4 T0294.try1-opt1.pdb.gz FOLDpro_TS1-scwrl ROBETTA_TS1-scwrl The C288, I290, and P292 residues do indeed correspond to the dimerization interface in 1gdhA, so we need to make a dimeric version of the protein and do any further polishing there. Wed Jun 7 19:10:35 PDT 2006 Kevin Karplus I have made a dimer based on try1-opt2 and the 1gdh[AB] template, and am optimizing it on cheep. Wed Jun 7 22:37:27 PDT 2006 Kevin Karplus dimer/try1 failed with an assertion failure before any improvements were made. It seems like one of the tertiary edges failed an OK() check. This probably means that there is a missing recompute_edge_lengths() somewhere. Thu Jun 8 08:55:58 PDT 2006 Kevin Karplus After bug fixes last night, the dimer/try1 run went fine. decoys/try1-opt2 resolved a number of clashes, but there are still some bad breaks. I will have to submit this morning to make the soft deadline, but more polishing is needed. Oops---I forgot to tell undertaker to optimize dimer/try1 as a dimer! I'll have to try again! (try2, with breaks and clashes turned up and constraints turned down, started on cheep). Thu Jun 8 11:13:16 PDT 2006 Kevin Karplus dimer/try2-opt1 does not seem to have reduced breaks much. The bad break before T237 is probably because of the helix through E236. Unwinding one or two residues from the helix would probably fix the problem. The E44-L45 break looks harder to fix---I might require reshaping S36-E44. The break at L144-W145 also looks tricky, as the both hydrophobic residues are currently exposed, which looks pretty wrong. The break at Q178-R179 is over packing. I may want to look at what Robetta did at these four points, and possibly do some cutting and pasting, since undertaker is having trouble closing the gaps. It may also be faster to work in the monomer. Thu Jun 8 13:55:37 PDT 2006 Kevin Karplus dimer/try2-opt2 still has terrible breaks: T0294.try2-opt2.pdb.gz breaks before (T0294)T237 with cost 8.43979 T0294.try2-opt2.pdb.gz breaks before (T0294)L45 with cost 4.19681 T0294.try2-opt2.pdb.gz breaks before (T0294)W145 with cost 3.18016 T0294.try2-opt2.pdb.gz breaks before (T0294)R179 with cost 2.56453 T0294.try2-opt2.pdb.gz breaks before (T0294)C147 with cost 1.79373 I'll try a polishing run on just the monomer, to see if that will help. If it does, I'll remake the dimer and repolish as a dimer. I'll include the dimer/try1-opt and dimer/try2-opt2 in the set of monomers that are being polished by try2 (running on lopez). ------------------------------------------------------------ Subject: CASP7 update - June 16 From: CASP Date: Fri, 16 Jun 2006 16:54:05 -0700 Bad news first ... It was brought to our attention that a paper was published on CASP7 structure T0294 ( http://dx.doi.org/10.1016/j.jmb.2006.05.018 ). As this happened less than 3 weeks after the target release, we are canceling this target according to our rules. No models will be assessed on this target. This information leak is not a fault of the SG center that provided the target, as the publication is from another group that solved the same protein. ------------------------------------------------------------