Wed Jun 4 09:14:03 PDT 2008 T0431 Make started Wed Jun 4 09:15:08 PDT 2008 Running on cheep.cse.ucsc.edu Wed Jun 4 09:16:22 PDT 2008 Kevin Karplus additional info: ph=6.5. Dimer in crystal assymmetric unit. Heme B bound. Why do they keep giving us huge dimers for the hand predictions? Homology model to 3b98A, 2iagA, 3b6hA (Cytochrome). I doubt that I'll be able to add anything to these 33%id over 490 residue templates. Thu Jun 5 04:09:03 PDT 2008 Kevin Karplus The try1 model and the top few alignments are all *very* similar. There are some bad breaks, so I should probably do polishing runs (without scwrling all the input alignments). try1-opt3 includes an ssbond (C27-C52) that seems to be in the alignments also, but the templates do NOT have ssbonds, even when they have close CYS residues, so I should turn off any bonuses for ssbonds! There is also no clustering of CYS residues in the templates, so even maybe_metal should be turned off. Before I do a polishing run, though, I think I should do another run from alignments without the maybe_ssbond cost function, to see if the disulfides appear without the ssbond cost function---it *does* look like SCWRL is introducing them in the alignments, So for try2, I am running from the alignments, but not SCWRLing the alignments as I usually do. If the try2 run comes up with ssbonds, I'll be very surprised (and more inclined to believe in their existence). Thu Jun 5 04:54:19 PDT 2008 Kevin Karplus This particular protein is a membrane-associated (endoplasmic reticulum) protein and goes by many names: Cytochrome P450 7A1 Cholesterol 7-alpha-monooxygenase CYPVII EC 1.14.13.17 Cholesterol 7-alpha-hydroxylase The heme binds ot the highly conserved C427. Thu Jun 5 12:29:23 PDT 2008 Kevin Karplus The try2 run has somewhat smaller gaps and clashes than the try1 run. I'll do a polishing run starting from both and trying to reduce clashes and breaks. Mon Jun 9 21:29:42 PDT 2008 Kevin Karplus MQAC1 is based mainly on Zhang-Server_TS4 MQAU1 is based mainly on Zhang-Server_TS1 With the try3 costfcn, the best models are MQAC1-opt3.gromacs0,repack-nonPC try3-opt3.gromacs0 MQAU1-opt3.gromacs0.repack-nonPC try2-opt3.gromacs0.repack-nonPC It looks like another polishing run is in order to reduce breaks and clashes further. All the models look fairly similar, with some slight differences in packing the helices. I'll do a polishing run for try4 from all models (including the metaserver models). Tue Jun 10 13:24:35 PDT 2008 Kevin Karplus try4 optimizes MQAU1-opt3.gromacs0 It looks pretty good, but I'm curious about whether the heme-binding site has been encroached on. I might want to add constraints to R430, P428, F432, S343. Thu Jun 19 11:58:26 PDT 2008 SAM-T08-MQAO hand QA T0431 Submitted Thu Jun 19 11:58:26 PDT 2008 SAM-T08-MQAU hand QA T0431 Submitted Thu Jun 19 11:58:26 PDT 2008 SAM-T08-MQAC hand QA T0431 Submitted Sun Jun 22 20:58:54 PDT 2008 Kevin Karplus Mapping the residues T0431 3b98A F420 W414 C427 C421 C427.CB-F420.CA 4.3 C427.CA-F420.C 4.8 P428 P422 P428.CD-F420.C 7.7 R430 F432 F426 F432.CB-C427.CB 9.4 S343 A335 S343.CB-C427.SG 9.5 F420.CA-S343.CB 7.3 R338 R330 F253 W245 F253.CG-C427.SG 11.9 I'll try adding these "heme" constraints to try5 and see if that improves anything. (The try2-opt3.repack-nonPC model statisfies the heme constraints best, so things probably have been disrupted a bit.) Mon Jun 23 09:41:19 PDT 2008 Kevin Karplus try5 improved the "heme" constraints, at some cost to the other cost functions. I think that the F253 constraint is wrong---the reesidue is about 23 Angstroms from C427, not 12, so is propbably not hte correct correspondence to W245 in 3b98A. Indeed, looking at the alignment, I see that the correspondence SHOULD have been T0431 3b98A L239 W245 L239.CG-C427.SG 11.9 There was also a typo I didn't catch: one of the F420s had been typed as F414. try5-opt3 is based mainly on try3-opt3.repack-nonPC, though the differences in the positions of the constrained residues is small between the different models. I should fix the heme constraints and turn them down to a more rational level for try6. Maybe add T0431 3b98A A268 V273 A268.O C427.SG 6.3 N272 N277 N272.CB C427.SG 7.3 N272.ND2 C427.SG 6.5 T273 A278 T273.CB C427.SG 9.0 I471 I465 I471.CG1 C427.SG 9.4 S341 A333 S341.CB C427.SG 10.0 I437 I431 I437.CD1 C427.SG 9.8 F83 Q94 no easy correspondence N107 K119 no easy correspondence Mon Jun 23 13:57:21 PDT 2008 Kevin Karplus try6 improved the heme constraints on try4, but not really to the point that try5 had. I turned them up for try7, to the point where try6 barely beats try5. It looks like I'll have 3 main models: try7 < try6 < try4 < MQAU1 < Zhang-Server_TS1 try5 < try3 < try2 < 3b6hA MQAC1 < Zhang-Server_TS4 Perhaps I should also do a run from just 3b98A and 2iagA, with the heme constraints, but not the sheet and helix constraints from the existing models. Mon Jun 23 17:35:52 PDT 2008 Kevin Karplus The try8 costfcn rates try8-opt3 very poorly, in large part because align_constraints are not well satisfied. I'll do a polishing run on try8, with align_constraints turned down and breaks and clashes turned up. Tue Jun 24 09:26:27 PDT 2008 Kevin Karplus The try9 costfcn likes best try7, try6, try5, try9 (all -opt3). The 4 different models I have now are try7 try5 try9 MQAC1 The core atoms agree to 0.5 Angstrom or better, but the RMSD between models for the whole chain is around 2-3 Angstroms. The biggest disagreements seem to be 170-235 and 81-119, all helices on the outside of the protein. Rosetta likes best try3-opt3.gromacs0.repack-nonPC, so maybe I should toss that into the mix also, though it should be similar to try5-opt3. Tue Jun 24 12:19:39 PDT 2008 Kevin Karplus I'll try to create one more meta-server model, optimized with the try9 costfcn, based on BAKER-ROBETTA models, which score the best of the complete server models with that cost function. Tue Jun 24 14:34:49 PDT 2008 Kevin Karplus try10-opt3.gromacs0.repack-nonPC scores a little worse than try9, but rosetta likes it better (almost as much as try3-opt3.gromacs0.repack-nonPC) Tue Jun 24 14:36:57 PDT 2008 Kevin Karplus Will submit ReadConformPDB T0431.try7-opt3.pdb < try6-opt3 < try4-opt3 < MQAU1-opt2.gromacs0 < Zhang-Server_TS1 ReadConformPDB T0431.try5-opt3.pdb < try3-opt3.repack-nonPC < try2-opt3.gromacs0.repack-nonPC < alignments(3b6hA?) ReadConformPDB T0431.try9-opt3.pdb < try8-opt3.gromacs0 < alignments(2iagA) ReadConformPDB T0431.try10-opt3.gromacs0.repack-nonPC.pdb < BAKER_ROBETTA_TS5 ReadConformPDB T0431.MQAC1-opt3.gromacs0.repack-nonPC.pdb < Zhang-Server_TS4