Thu Jun 19 09:53:50 PDT 2008 T0462 Make started Thu Jun 19 09:54:07 PDT 2008 Running on peep.cse.ucsc.edu Thu Jun 19 10:12:37 PDT 2008 Kevin Karplus BLAST finds homologs in PDB: 2gcxA (E-value 3e-07) 2h3jA (E-value 4e-05) so I suspect that this is a fairly easy homology model. Thu Jun 19 12:11:29 PDT 2008 Kevin Karplus The HMMs like template length E-value SCOP 2h3jA 76 1.4820e-11 2gcxA 76 2.0172e-11 1fx7A 231 1.1065e-04 a.4.5.24,a.76.1.1,b.34.1.2 60088,60089,60090 2dtrA 215 1.8409e-04 a.4.5.24,a.76.1.1,b.34.1.2 16193,18399,24450 1qw1A 122 5.2697e-04 2qq9A 213 1.1213e-03 Conserved residues are mainly in the first 75, and that is where alignments to 2h3jA, 2gxcA, 1fx7A, 2dtrA, 1qw1A, and 2qq9A are, so the hard part will be the second domain. Maybe I should try a subdomain prediction for G75-K154 Thu Jun 19 12:28:11 PDT 2008 Kevin Karplus G75-K154 is looking like an ORFan. We may want to conjecture that that region is a domain like one of the other domains in 1fx7A, 2dtrA, ..., but that should fall out of any global alignments tried for those templates. I just hope they have NMR data for the second half of the protein, and that I don't go to a lot of work predicting the structure for something that is unstructured or for which data wasn't collected. This is supposed to be an iron-transport protein, so I should look for an iron-binding site. Thu Jun 19 15:01:57 PDT 2008 Kevin Karplus The G75-K154 domain also has 2gcxA as a top hit (though with a poor E-value). So this looks like 2 copies of the same domain. I'll make a chimera of try1-opt3 and G75-K154/try1-opt3 and put together a costfcn that has the good parts of both. Thu Jun 19 15:22:17 PDT 2008 Kevin Karplus The chimera-try1-C1 does not have a good placement of the two domains (crossing over between I81 and L82), but I'll try optimizing with the try2 costfcn (that takes sheets and helices from both) and see what comes up. I should probably do a subdomain for the first domain by itself also: M1-A83 Thu Jun 19 18:21:05 PDT 2008 Kevin Karplus The try2 run finished and scores much better than the try1 run, even though it still has fairly bad clashes and breaks. The packing of the two domains against each other is pretty arbitrary, but does not seem obviously wrong. At least W139 is buried. Thu Jun 19 18:31:54 PDT 2008 Kevin Karplus I should make a chimera using M1-A83/try1-opt3 for residues M1-V74 and try2-opt3 for the rest, then optimize that. Thu Jun 19 21:15:22 PDT 2008 Kevin Karplus try3-opt3 still needs a little gap closing, but otherwise looks pretty good. I don't think the monomers are packed against each other right, but it might be hard to move them. I'm not sure how they should be packed. Mon Jun 23 11:55:20 PDT 2008 Kevin Karplus MQAU favors Zhang-Server MQAC favors FALCON_CONSENSUS Mon Jun 23 12:58:46 PDT 2008 Kevin Karplus The MQAC3-opt3 model is based mainly on PS2-server_TS3 The MQAU3-opt3 model is based mainly on Pcons_dot_net_TS5 Mon Jun 23 13:05:17 PDT 2008 Kevin Karplus Apparently I had don a try4 polishing run to polish try3 without making any notes on it in the README file. Re-scoring models with try4.costfcn, it favors MQAU3-opt3 try4-opt3.gromacs0 try3-opt3 MQAC3-opt3.gromacs0 try2-opt3.gromacs0 The MQAC3-opt3 model is not well-packed. Perhaps I need to select among the models without favoring the try3 sheets so much. I'll do meta_MQAU1 and meta_MQAC1 runs witth the try1 costfcn also. Mon Jun 23 13:17:28 PDT 2008 Kevin Karplus The MQAU3-opt3 run places the two barrels differently from try4. I wonder if I should make a chimera that uses that placement, but replaces the domains with the optimized barrels from the subdoamins. Mon Jun 23 14:17:16 PDT 2008 Kevin Karplus The MQAU1-opt3 run is based mainly on SAM-T08-server_TS1 The MQAC1-opt3 run is based mainly on Zhang-Server_TS2 Mon Jun 23 18:00:31 PDT 2008 Kevin Karplus MQAU1-opt3.gromacs0 scores best with the try4 costfcn. Wed Jun 25 09:05:09 PDT 2008 Kevin Karplus I accidentally redid the MQA computations, because the CASP people had re-released the server tarball. The metaserver run was not redone, since everything was judge "up to date" (no dependency on the server files, I guess). Sat Jun 28 14:40:53 PDT 2008 SAM-T08-MQAO hand QA T0462 Submitted Sat Jun 28 14:40:54 PDT 2008 SAM-T08-MQAU hand QA T0462 Submitted Sat Jun 28 14:40:54 PDT 2008 SAM-T08-MQAC hand QA T0462 Submitted Fri Jul 4 09:45:33 PDT 2008 Kevin Karplus My models all seem to have 2 barrel domains, but the main difference between models seems to be how the barrels are packed against each other. There may not be much point to making more models at this, point---trying to decide on the packing is probably most useful. I think I need to do break and clash reduction on try4 before I can fairly compare it with MQAU3. try5 uses the sheet and helix constraints from try4, MQAU1, and MQAU3, plus sheet constraints from the try1 runs of the separate domains. I'll polish try4, MQAU3, and MQAU1 separately with the same costfcn for try5, try6, and try7. Fri Jul 4 12:26:25 PDT 2008 Kevin Karplus Of the polishing runs, the order is try6 < MQAU3-opt3 try5 < try4-opt3 try7 < MQAU1-opt3 The try6 and try7 models seem quite similar, so I'll just keep the better one (try6). In try6, I don't like the way that L80-A83 has been wound into a helix (predicted to be strand). It is straight, but not attached to a sheet in try5-opt3. I think I like try5 a bit better than try6, even though it scores slightly worse with a "fair" costfcn that takes sheets and helices from both. I'll try polishing try5 to close gaps and see how it scores then. Fri Jul 4 17:12:46 PDT 2008 Kevin Karplus OK try8 not scores better than try6, both with the try8 costfcn and the "fair" costfcn. Fri Jul 4 17:41:42 PDT 2008 Kevin Karplus I submitted with the following comment: For this target, I did optimization of the whole chain, and also briefer runs with two overlapping subdomains: M1-A83 and G75-K154. Some of the models I'm submitting are metaserver models optimized from the server models, and others are from the two subdomains optimized separately then patched together to make a whole protein. Model 1 T0462.try8-opt3.pdb # < try5-opt3 < try4-opt3 < try3-opt3 < chimera-M1-try2 chimera-M1-try2: M1-V74 is from M1-A83/T0462.try1-opt3.gromacs0.repack-nonPC.pdb G75-K154 is from T0462.try2-opt3.gromacs0.repack-nonPC.pdb 2 T0462.try6-opt3.pdb # < MQAU3-opt3 < Pcons_dot_net_TS5 3 T0462.MQAC3-opt3.pdb # < PS2-server_TS3 4 T0462.try2-opt3.gromacs0.pdb # < chimera-try1-C1 chimera-try1-C1: M1-I81 from T0462.try1-opt3.pdb L82-K154 from G75-K154/T0462.try1-opt3.pdb try1-opt3 < align(1b8wA? 1fx7A?) 5 T0462.MQAC1-opt3.gromacs0.pdb # < Zhang-Server_TS2