Fri May 30 09:11:51 PDT 2008 T0425 Make started Fri May 30 09:19:06 PDT 2008 Running on cheep.cse.ucsc.edu Fri May 30 09:19:52 PDT 2008 Kevin Karplus BLASTing PDB finds no strong hits (2gciA at E-value 0.37). Fri May 30 09:54:36 PDT 2008 Kevin Karplus The t06 iterated search includes 1jwqA and 1xovA in the multiple sequence alignment as essentially full-length matches, so I suspect that this will be easy fold-recognition modeling. Fri May 30 14:34:45 PDT 2008 Kevin Karplus There are bad breaks before K96 and I164, but otherwise the try1-opt3 model looks pretty good. I'll do a polishing run with breaks and clashes turned up. It looks like H11, E25, H77, E144 form an active site (N79 may also be involved). I should look at the homologs (1jwqA and 1xovA, c,56,5,6) to see whether there are some constraints I should copy. 1jwqA has a zinc coordinated by the first three of those residues (H10=H11, E26=E25, H80=H77), giving distances: H77.ND1 H11.NE2 3.228 H11.NE2 E25.OE1 3.024 H77.ND1 E25.OE1 3.083 The other side of the zinc has bound water. The other conserved GLU has distances E144.OE1 H77.ND1 4.328 E144.OE2 E25.OE1 4.747 The N79 may also be involved N79.ND2 E25.OE1 7.411 N79.ND2 H77.ND1 8.291 From 1xovA, I get the zinc-binding distances H77.ND1 H11.NE2 3.191 H11.NE2 E25.OE2 3.132 H77.ND1 E25.OE2 3.153 The other side of the zinc is covered by backbone atoms, rather than by water. Fri May 30 15:04:22 PDT 2008 Kevin Karplus I put some constraints in for the zinc site for try2 and am polishing the try1 models. Fri May 30 18:30:35 PDT 2008 Kevin Karplus try2-opt3 looks pretty good. There are some minor breaks and clashes that I'll try to fix in try3. Fri May 30 19:49:29 PDT 2008 Kevin Karplus try3-opt3 looks pretty good. I think this is done (at least until I see the metaserver results). On second thought, some of the solvated sidechains look a little weird in try3-opt3. Maybe I should add some salt bridges: R117.NH2 D153.OD1 K55.NZ D27.OD1 K123.NZ D120.OD2 H11.ND1 D15.OD2 Fri May 30 20:06:19 PDT 2008 Kevin Karplus Started optimization with salt bridges as try4. Fri May 30 21:15:12 PDT 2008 Kevin Karplus Differences between models try1 through try4 are tiny and consist mainly of sidechain changes for exposed residues, which may not even be well-defined in the crystal. I think I'm done with this, though I'll look at it again when the metaserver results are in. Sun Jun 8 11:06:29 PDT 2008 SAM-T08-MQAO hand QA T0425 Submitted Sun Jun 8 11:06:29 PDT 2008 SAM-T08-MQAU hand QA T0425 Submitted Sun Jun 8 11:06:29 PDT 2008 SAM-T08-MQAC hand QA T0425 Submitted Sun Jun 8 19:52:47 PDT 2008 Kevin Karplus The MQAC quality assessment favors the Zhang-Server The MQAU quality assessment favoersSAM-T08-server, then the Zhang-Server. Sun Jun 8 22:13:09 PDT 2008 Kevin Karplus Started metaserver runs with try3 as costfcn. Mon Jun 9 12:10:59 PDT 2008 Kevin Karplus costfcn favored models try2 try4-opt3 try3-opt3 try2-opt3 MQAC1-opt3 MQAU1-opt3 try3 try4-opt3 try3-opt3 try2-opt3 MQAC1-opt3 MQAU1-opt3 try4 try4-opt3 try3-opt3 try2-opt3 MQAC1-opt3 MQAU1-opt3 MQAC1-opt3 mainly from HHpred2_TS1 MQAU1-opt3 mainly from HHpred2_TS1 The metaserver models are almost indistinguishable from the models generated by SAM and undertaker. I'll submit ReadConformPDB T0425.try4-opt3.pdb ReadConformPDB T0425.try3-opt3.gromacs0.repack-nonPC.pdb # best Rosetta energy ReadConformPDB T0425.try2-opt3.pdb ReadConformPDB T0425.MQAC1-opt3.pdb ReadConformPDB T0425.MQAU1-opt3.pdb Mon Jun 9 12:26:25 PDT 2008 Kevin Karplus So submitted. Mon Jul 14 16:12:00 PDT 2008 Kevin Karplus PDB released SAM-T08-server seems to be in the middle of the pack of predictions, with a GDT of 67% (top in real_cost is Zhang-Server_TS3 at 72.8%) The best model I submitted was model 4 (GDT 71.27%), about rank 125. The best model I created was MQAC1-opt1-scwrl (GDT 70%, rank 72). Not an impressive achievement, even for a metaserver.