Wed Jul 12 14:09:05 PDT 2000 T0111 Enolase, E. coli This is definitely a homology-modeling target, with blast finding 17 PDB chains that are excellent hits: 1ebg[AB] 1ebh[AB] 1one[AB] 2one[AB] [34567]enl 1nel 1els 1pdy 1pdz (Double-blast finds a few more remote homologs that blast does not: 1mra, 2mnr, 1mns, 1dtn, 1mdr, 1mdl, but we're not very interested in these for this application.) Since we know this is a dimer, we probably want to use one of the four 1ebg[AB] 1ebh[AB] 1one[AB] 2one[AB] all of which have identical scores, so probably identical chains. The FSSP rep for all is 4enl. 1one has the best resolution (at 1.8 Ang), 2one is the newest, and is within 1.0Ang of 1one for C-alpha (not surprising, since it used 1one as a starting point in molecular replacement). I suspect we'd do best using 2oneAB as our starting point. The target has some insertions and deletions relative to the template (according to the blast alignment), but I doubt that we will do much better than blast at aligning this close a homolog. The T0111.t2k_1.a2m file has 3 deletions relative to 2oneA and 4 insertions, all very close to where BLAST does similar things. NOTE: We should grab the 2ry structure from 2oneA, and NOT use the neural net prediction! We may want to fuss around a bit with where the gaps go, and we may want to try SCWRL or even undertaker to fiddle with the sidechains and close the gaps. Sat Aug 26 14:41:29 PDT 2000 Remade 2track predictions The top scores are overwhelmingly 2oneA, 1oneA, 4enl, as expected. We may want to look at the 2track alignments to see if they are any better. Tue Sep 5 10:35:22 PDT 2000 remaking 2track Top alignments 2oneA/T0111-2oneA-global 2oneA 436 -430.66 -449.87 1.3e-195 4 insertions and 5 deletions only one not in coil 4enl/T0111-4enl-global 4enl 436 -430.45 -449.42 3.0e-195 identical insertions and deletions to 2oneA/T0111-2oneA-global 1oneA/T0111-1oneA-global 1oneA 436 -430.66 -449.87 3.0e-195 identical insertions and deletions to 2oneA/T0111-2oneA-global 4enl/4enl-T0111-global T0111 431 -414.33 -425.79 8.0e-185 3 gaps change by 1 position and one gap changes by 2 positions. 2oneA/2oneA-T0111-global T0111 431 -413.11 -423.34 4.2e-184 1oneA/T0111-1oneA-vit 1oneA 436 -430.04 -423.03 5.9e-184 2oneA/T0111-2oneA-vit 2oneA 436 -429.98 -422.99 6.0e-184 2oneA/T0111-2oneA-local 2oneA 436 -434.49 -422.31 1.2e-183 4enl/T0111-4enl-local 4enl 436 -434.33 -422.14 1.6e-183 1oneA/T0111-1oneA-local 1oneA 436 -434.54 -422.35 1.6e-183 4enl/T0111-4enl-vit 4enl 436 -429.82 -422.82 1.6e-183 1oneA/1oneA-T0111-global T0111 431 -410.94 -417.40 2.4e-181 2oneA/2oneA-T0111-vit T0111 431 -412.72 -409.13 6.2e-178 4enl/4enl-T0111-vit T0111 431 -413.98 -408.72 1.9e-177 4enl/4enl-T0111-local T0111 431 -417.99 -405.72 3.9e-176 2oneA/2oneA-T0111-local T0111 431 -416.63 -404.18 8.7e-176 1oneA/1oneA-T0111-vit T0111 431 -411.12 -403.76 2.9e-175 1oneA/1oneA-T0111-local T0111 431 -414.34 -400.15 5.7e-174 2oneA/T0111-2oneA-2track-global 2oneA 436 -332.37 -339.99 1.8e-147 1oneA/T0111-1oneA-2track-global 1oneA 436 -325.46 -337.40 1.3e-146 4enl/T0111-4enl-2track-global 4enl 436 -321.40 -331.16 5.3e-144 2oneA/T0111-2oneA-2track-local 2oneA 436 -336.73 -310.29 7.0e-135 1oneA/T0111-1oneA-2track-local 1oneA 436 -331.15 -303.34 7.7e-132 4enl/T0111-4enl-2track-local 4enl 436 -327.08 -301.65 5.7e-131 4enl/4enl-T0111-fssp-global T0111 431 -151.43 -113.49 2.5e-49 moves mid-strand insert to end of strand. unaligns a residue next to insert unaligns several residues around big deletion (perhaps too many?) moves insert to other end of a beta strand Made choices on the insert positions and saved as 4enl-T0111-karplus.a2m Probably should transfer to 2oneA though, since that is slightly closer homolog (according to our models and to blast). Looked at 2oneA/T0111-2oneA-2track-global, and modified it based on 4enl modifications. USE 2oneA/T0111-2oneA-karplus.a2m Fri Sep 8 10:26:03 PDT 2000 Kevin Karplus First undertaker run got wedged, and had to be killed after 24 hours (most of which were spent doing nothing) t111-2oneA.0.10.pdb ok 19.1603 bits/residue, 57 clashes 43.2838 breaks t111-2oneA.1.10.pdb ok 19.0682 bits/residue, 42 clashes 49.3711 breaks t111-2oneA.2.10.pdb destroys barrel t111-2oneA.3.10.pdb destroys barrel t111-2oneA.4.10.pdb ok 19.1014 bits/residue, 23 clashes 62.1585 breaks t111-2oneA.5.10.pdb ok 19.3791 bits/residue, 43 clashes 59.3179 breaks t111-2oneA.6.10.pdb destroys barrel Sun Sep 10 11:07:38 PDT 2000 Starting try2 from try1/t111-2oneA.0.10.pdb For try2, several of the optimization iterations unfold, but t111-2oneA.2.10.pdb is ok and is slightly better than the starting point. t111-2oneA 18.4357 bits/residue, 57 clashes t111-2oneA.2.10.pdb 18.2445 bits/residue, 53 clashes 0.000190186 breaks We might want to try again, with a higher dry_spot_weight and lower wet_spot_weight. The very polar face of the barrel may be making it tear apart in the scoring function. try3 makes a slight improvement on the first superiteration, t111-2oneA 21.7207 bits/residue, 53 clashes t111-2oneA-scwrl 20.9659 bits/residue, 52 clashes t111-opt.0.10.pdb 20.8655 bits/residue, 47 clashes but the run was terminated to allow finishing t128. Try again after t128 submitted. Mon Sep 11 10:01:37 PDT 2000 try4 also terminated after one superiteration. Try5 completed without unfolding. try5/t111-opt 20.3156 bits/residue, 33 clashes Try6 completed without unfolding try6/t111-opt-scwrl-rot 18.5752 bits/residue, 34 clashes The final beta strand has come unattached. This broken strand is already noticeable at try5, try4, try3, try2, but was ok in try1/t111-2oneA.0.10.pdb So do we submit try1/t111-2oneA.0.10.pdb, which has lots of clashes and breaks, but gets the beta strand or try6/t111-opt-scwrl-rot which scores better, has fewer clashes, but loses that beta strand? The predicted structure is a pair of 2-residue strands (rather than the 3-residue strands in 2oneA, and the proline at the beginning of the final strand may disrupt it. Maybe I should make try6 my first model and try1 my second.