Mon Jun 26 09:46:00 PDT 2006 T0346 Make started Mon Jun 26 09:46:25 PDT 2006 Running on cheep.cse.ucsc.edu BLAST gets excellent full-length hits to 2gw2A (72% gapless over 170 residues) and 1a58 (62% gapless over 165 residues). Mon Jun 26 10:14:51 PDT 2006 Kevin Karplus Interestingly, the t04 and t06 alignments do not appear to have 2gw2A, though there are several other peptidyl-prolyl isomerase G proteins. Ah---it seems that 2g2wA is still on publication hold, so we have the sequence in dunbrack-pdbaa, but the structure is not yet available. Interestingly, pdb says that the sequence is not available, but it is in dunbrack-pdbaa. At least the 1a58 cyclophilin is available. Mon Jun 26 16:51:54 PDT 2006 Kevin Karplus The HMMs like best 1yndA, 1dywA, 2bitZ, 1ihgA, 1xo7A, ... (b.62.1.1, cyclophilin) These are all so close that even the sidechain placements are coming out the same. A 1a58 alignment is what try1 ended up optimizing. I think this is a polish-and-be-done model. Not quite: we need to make a dimer. 1ynd[AB] would probably be an appropriate template to base it off of. I made 1yndA/T0346-1yndA.dimer-a2m from the str2-near-backbone local alignment. Mon Jun 26 17:27:59 PDT 2006 Kevin Karplus Before making the dimer, I'll polish up the monomer, though there is not much to do: a few mild clashes and minor breaks. Mon Jun 26 21:58:02 PDT 2006 Kevin Karplus try2-opt2 reduces clashes and breaks substantially, though rosetta disagrees, liking try1-opt2 a tiny bit better for repacking. Mon Jun 26 22:20:57 PDT 2006 Kevin Karplus I can hardly see the difference between try1-opt2 and try2-opt2, but I'll do one more polishing run with no constraints. Mon Jun 26 23:57:42 PDT 2006 Kevin Karplus try3-opt2 is definitely into the realm of diminishing returns. I just need to make a dimer of it and optimize the dimer. Tue Jun 27 10:41:03 PDT 2006 Kevin Karplus I made a dimer from try3-opt2, using the 1ynd[AB] alignment in 1yndA/T0346-1yndA.dimer-a2m as a template. I'll optimize it (without constraints) to improve packing. dimer/ try1 started on orcas Tue Jun 27 11:12:41 PDT 2006 Kevin Karplus The initial dimer was very loose, and we have predicted burial in the interface, so we may have to force things closer together, if the dimer/try1 run doesn't manage to do that. Possibilities include increasing phobic_fit and dry12 terms, reducing the soft_clashes penalty, and pushing the monomers closer together with ProteinShop. Let's see what undertaker can manage on its own first. Tue Jun 27 12:59:19 PDT 2006 Kevin Karplus dimer/try1-opt1 is still pretty open in the center. Tue Jun 27 20:20:32 PDT 2006 Kevin Karplus dimer/try1-opt2 still has a lot of water in the center---we need to push the monomers closer together. Wed Jul 12 22:43:08 PDT 2006 Kevin Karplus The runs here were old enough not to have gromacs0.repack-nonPC models built, so I'm building them now. Wed Jul 12 23:03:28 PDT 2006 Kevin Karplus Rosetta liked try1-opt2.gromacs0.repack-nonPC best, though all models were close. It might be worthwhile to do a polishing run from just the gromacs0.repack-nonPC models, as that has worked well at improving scores in other comparative modeling situations. Wed Jul 12 23:08:49 PDT 2006 Kevin Karplus try4 polishing started on lopez. Wed Jul 12 23:11:27 PDT 2006 Kevin Karplus I'm not going to worry about the big "hole" in the dimer---in 1ynd there is a large co-factor there: SANGLIFEHRIN A C_60 H_91 N_5 O_1 When try4 has finished, I'll dimerize it and polish the dimer, and that will be it. Thu Jul 13 00:42:01 PDT 2006 Kevin Karplus try4-opt2 does not score quite as well as try3-opt2 with the try4 costfcn, but Rosetta likes decoys/T0346.try4-opt2.gromacs0.repack-nonPC.pdb.gz a little bit better than try1-... which was its previous favorite. Thu Jul 13 09:42:05 PDT 2006 Kevin Karplus I looked at the second-highest hit (1dywA) and saw that it dimerizes differently (with the binding pockets exposed, so I'll make a dimer using that. Thu Jul 13 09:50:10 PDT 2006 Kevin Karplus dimer-try4-1dywA has a bit of a clash, but I'll try optimizing it away in dimer/try2 (started on cheep). Thu Jul 13 09:54:57 PDT 2006 Kevin Karplus Yet another dimerization possiblity is 2rmc[AC] Thu Jul 13 10:35:12 PDT 2006 Kevin Karplus dimer/try3 started on abyss to optimize dimer-try4-2rmcA Thu Jul 13 12:25:06 PDT 2006 Kevin Karplus dimer-try2 changed the dimerization completely from 1dywA. Quite frankly, I think that without a co-factor, the dimerization is pretty random. Thu Jul 13 13:50:45 PDT 2006 Kevin Karplus I'm ready to submit the dimers: T0346.try3-opt2.unpack.gromacs0.repack-nonPC # from 2rmc T0346.try1-opt2.unpack.gromacs0.repack-nonPC # from 1ynd dimer-try4-1dywA T0346.try2-opt2.unpack.gromacs0.repack-nonPC # mangled from 1dyw dimer-try4-1yndA I'm also ready to submit the monomers: T0346.try4-opt2.gromacs0.repack-nonPC.pdb T0346.try3-opt2.pdb T0346.try2-opt2.pdb T0346.try1-opt2.pdb T0346.undertaker-align.pdb model 1 # from 1yndA Thu Jul 13 14:27:24 PDT 2006 Kevin Karplus Both monomer and dimer submissions done. Mon Aug 21 15:20:01 PDT 2006 Kevin Karplus Our best (monomer) model is try1-opt2.repack-nonPC, with GDT 98.4%, 0.5378 CA RMSD, 1.1971 all-atom RMSD. The best we submitted was model4, try1-opt2, again GDT 98.4%, 0.5378 CA RMSD, but all-atom RMSD=1.3758. Note: this was the easiest target of the first 58 evaluated (all that are available so far).