Wed Jun 11 09:24:20 PDT 2008 T0446 Make started Wed Jun 11 09:24:37 PDT 2008 Running on cheep.cse.ucsc.edu Wed Jun 11 11:12:21 PDT 2008 Kevin Karplus This looks like a homology model to 3b77A (not in SCOP) with no other close hits. VAST reports sequence matches to pfam08000: DUF1696 and structural matches to 2hthB, which is a b.55.1.12 domain. The b.55.1, PH-domain-like, superfamily seems to be responsible for a number of the top hits, though only 3b77A ever scores well. Wed Jun 11 11:49:22 PDT 2008 Kevin Karplus The first 4 undertaker-align models superimpose fairly well and cover roughly residues P25-L124. Because the top template scores so well, the alignment constraints should be reasonably weighted and not contaminated by the really weak distant predictions. It looks like try1 is picking stuff up from other templates also: 3b77A, 2kfnA, 1jr7A, 2bsqA. I don't know if the other templates are full-length matches or just short fragments. 1jr7A has some fairly long matches and is b.82.2.3, not b.55.1.*. My guess is that these extra templates were just providing a strand or a slightly larger super-secondary element. Wed Jun 11 12:04:12 PDT 2008 Kevin Karplus The burial and contact prediction look pretty good in try1-opt1 and try1-opt2. The secondary structure looks ok, except for strand L33-V39, which is weakly predicted to be either helix or strand with helix slightly stronger. Wed Jun 11 12:39:03 PDT 2008 Kevin Karplus The try1-opt3 model looks fairly good. I'll try a polishing run to reduce breaks and clashes. Wed Jun 11 15:03:43 PDT 2008 Kevin Karplus try2-opt3 reduced clashes, but still has bad breaks: T0446.try2-opt3.pdb.gz breaks before (T0446)A73 with cost 4.33479 T0446.try2-opt3.pdb.gz breaks before (T0446)D29 with cost 2.84634 I'll do another polishing run with breaks turned up higher. Wed Jun 11 17:28:04 PDT 2008 Kevin Karplus try3-opt3 improved scores a little, cutting breaks in half, but more break and clash removal still needed. Maybe I should do an optimization just from the gromacs-optimized models. Wed Jun 18 19:03:59 PDT 2008 Kevin Karplus MQAC favors the Zhang-Server models MQAU favors SAM-T08-server, METATASSER, and Zhang-Server I'll start the two metaserver runs (with try1.costfcn) on the moai cluster. I may want to do metaserver runs with other costfcns also. Thu Jun 19 12:00:22 PDT 2008 SAM-T08-MQAO hand QA T0446 Submitted Thu Jun 19 12:00:22 PDT 2008 SAM-T08-MQAU hand QA T0446 Submitted Thu Jun 19 12:00:22 PDT 2008 SAM-T08-MQAC hand QA T0446 Submitted Sun Jun 29 20:37:20 PDT 2008 Kevin Karplus The MQAC1 and MQAU1 runs failed, because the SAM-T08-MQA*.read_under files had not been created. I'm re-running them. Mon Jun 30 10:48:46 PDT 2008 Kevin Karplus MQAC1 scores well with the try4 costfcn, and is based on BAKER-ROBETTA_TS2 MQAU1 does not score as well and is based on SAM-T08-server_TS1 The biggest advantage of MQAC1 is that inserts a strand L7-A11 into the edge of the sandwich. making it more barrel-like. This isn't really compatible with the secondary structure prediction, though. The template (3b77A) does have that extra strand, though, so it is a promising alternative. I'll do a polishing run with the MQAC1 sheets. Mon Jun 30 11:04:05 PDT 2008 Kevin Karplus try5 started on peep. Mon Jun 30 17:14:11 PDT 2008 Kevin Karplus try5-opt3.gromacs0 scores best with try5 costfcn, so more polishing probably needed. It does indeed improve on MQAC1 model. Mon Jun 30 22:19:06 PDT 2008 Kevin Karplus I tried making a chimera of try5-opt3.gromacs0 and try4-opt3: M1-I26 from T0446.try5-opt3.gromacs0.pdb M27-L124 from try4-opt3.pdb I'll try optimizing that chimera as try6 Tue Jul 1 06:46:25 PDT 2008 Kevin Karplus try6 doesn't score quite as well as try5, but rosetta likes try6-opt3.gromacs0.repack-nonPC best. Note: try6 lost the extra strand of try5, though it was present in th chimera! Perhaps I should make another attempt at removing the clashes and breaks of the chimera, with somewhat lower clash and break penalties and higher weight on the crucial strand. Tue Jul 1 09:30:57 PDT 2008 Kevin Karplus With the try7 costfcn, try7-opt3.gromacs0 scores almost as well as try5-opt3. I think I'd like to try a chimera-try7-try5, that takes D55-I74 from try5-opt3 and the rest from try7-opt3.gromacs0 Tue Jul 1 09:46:34 PDT 2008 Kevin Karplus Optimizing chimera-try7-try5 as try8. Tue Jul 1 14:01:14 PDT 2008 Kevin Karplus try8 is not very different from previous models. I think I'm spinning my wheels now. I'll submit ReadConformPDB T0446.try8-opt3.pdb # < chimera-try7-try5 chimera-try7-try5: mostly from T0446.try7-opt3.gromacs0.pdb D55-I74 from T0446.try5-opt3.pdb try7 < chimera-try5-try4 chimera-try5-try4: M1-I26 from T0446.try5-opt3.gromacs0.pdb M27-L124 from try4-opt3.pdb ReadConformPDB T0446.try6-opt3.gromacs0.repack-nonPC.pdb # < chimera-try5-try4 ReadConformPDB T0446.try5-opt3.pdb # < MQAC1-opt3.gromacs0 < BAKER_ROBETTA_TS2 ReadConformPDB T0446.try4-opt3.pdb # < try3-opt3.gromacs0 < try2-opt3 < try1-opt3 < align(3b77A? 2bsqA?) ReadConformPDB T0446.MQAU1-opt3.gromacs0.repack-nonPC.pdb # < SAM-T08-server_TS1