Fri Jun 13 09:24:58 PDT 2008 T0451 Make started Fri Jun 13 09:25:18 PDT 2008 Running on cheep.cse.ucsc.edu Fri Jun 13 09:54:16 PDT 2008 Kevin Karplus The t06 alignment contains 22 PDB chains, mostly ketosteroid isomerases. The t2k alignment has only 2: 1nwwA and 1nu3A, both Limonene-1,2-Epoxide Hydrolase (which are in the t06 alignment also). Fri Jun 13 11:32:11 PDT 2008 Kevin Karplus Top hits for T0451 are all d.17.4.* superfamily, NTF2-like (fold: cystatin-like), but from different families. The more specific t2k search favors the YesE family (d.17.4.10), 1s5aA. with Limonene-1,2-epoxide hydrolase (d.17.4.8), 1nwwA second. The broader t06 search favors Hypothetical protein egc068 from a soil-derived mobile gene cassette (d.17.4.11) 1tuhA followed by Ketosteroid isomerase-like (d.17.4.3) 2a15A. So we have a clear superfamily, but at least 4 families in strong contention. The models in undertaker-align.pdb all seem to be very close to each other, so I think that this will turn out to be a fairly easy homology model. Fri Jun 13 14:15:34 PDT 2008 Kevin Karplus The try1-opt3 model looks fairly good, except for detaching strand G51-G55 I should add the align1 sheet constraints: SheetConstraint (T0451)V108 (T0451)V114 (T0451)F126 (T0451)C120 hbond (T0451)A109 34.7121 SheetConstraint (T0451)F40 (T0451)F44 (T0451)C120 (T0451)E124 hbond (T0451)W43 34.7121 SheetConstraint (T0451)D113 (T0451)R115 (T0451)C120 (T0451)K118 hbond (T0451)R115 34.7121 SheetConstraint (T0451)T87 (T0451)F99 (T0451)V114 (T0451)K102 hbond (T0451)V89 34.7121 SheetConstraint (T0451)S69 (T0451)T72 (T0451)L100 (T0451)G97 hbond (T0451)H71 34.7121 SheetConstraint (T0451)Q75 (T0451)S78 (T0451)G97 (T0451)R94 hbond (T0451)I76 34.7121 SheetConstraint (T0451)D80 (T0451)S84 (T0451)F93 (T0451)V89 hbond (T0451)R81 34.7121 SheetConstraint (T0451)F40 (T0451)F44 (T0451)G55 (T0451)G51 hbond (T0451)F42 34.7121 to the try1-opt3.sheets for try2 and try remaking from alignments. Fri Jun 13 14:30:37 PDT 2008 Kevin Karplus So started on cheep. Fri Jun 13 18:33:48 PDT 2008 Kevin Karplus try2-opt3 scores better than try1-opt3, but gromacs0 improves it, reducing both clashes and breaks. I should do a polishing run with breaks and clashes turned up. Fri Jun 13 21:19:39 PDT 2008 Kevin Karplus try3-opt3 is fairly tight, with 3 small breaks. Running through gromacs closes those breaks but opens another. It may be worth doing another polishing run starting from just the models that have been run through gromacs to get the clashes and breaks even smaller. Hmm, it looks like the strand pair SheetConstraint S78 D80 R94 E92 hbond S78 has gone missing in try3---the residues F70-S77 don't straighten out to form a strand either. I'm not sure now what to do for try4---should I try to fix the strand, or just pack things tighter? Part of the problem is that F70-I76 seems to have more residues than in the templates aligned to, and figuring out where to put these residues is a problem. Tue Jun 17 13:14:53 PDT 2008 Kevin Karplus SAM-T08-MQAU likes the SAM-T08-server_TS1 model best (predicted GDT 73.2%), then pro-sp3-TASSER (TS2 and TS3), then Zhang-Server (TS 2,3,1) SAM-T08-MQAC likes Zhang-Server models, METATASSER and pro-sp3-TASSER The SAM-T08-server models is much lower (rank 25, predicted GDT 68.8% instead of 73.1% for the top Zhang-Server model). I'll do the metaserver runs with the try1 costfcn. Tue Jun 17 14:57:03 PDT 2008 Kevin Karplus MQAC1 works primarily off of Zhang-Server_TS1 MQAU1 works primarily off of SAM-T08-server_TS1 The try1 costfcn likes try3-opt3 MQAC1-opt3.gromacs0 MQAU1-opt3 try2-opt3 try1-opt3 try2 likes try3-opt3 MQAC1-opt3.gromacs0 try2-opt3 MQAU1-opt3 try1-opt3.gromcas0.repack-nonPC try3 likes try3-opt3 try2-opt3.gromacs0.repack-nonPC MQAC1-opt3.gromacs0 try1-opt3.gromcas0.repack-nonPC MQAU1-opt3.gromacs0 The biggest differences between try3 and MQAC1 (from Zhang-Server_TS1) are in the N- and C-terminal tails, but there are also differences in the edge strands also. I think that I like the more regular strands of the MQAC1 model better, so I'll do an optimization from all models that uses those sheet constraints. Hmm, that won't work, because both try3 and try2 have much smaller clashes and breaks. I'll have to polish MQAC1 first. Tue Jun 17 19:00:21 PDT 2008 Kevin Karplus OK, try4-opt3 scores almost as well as try3-opt3 and gets the MQCA1 sheets and helices better than MQAC1 does. Rosetta likes it almost as well as try3 also. I like try3-opt3 better up to about E38, but try4 for the rest. Let me try making a try3-try4 chimera and optimizing it. Tue Jun 17 19:40:45 PDT 2008 Kevin Karplus The chimera-try3-try4 model has some pretty nasty clashes, and it doesn't score as well as I expected on some of the other measures either, but I'll try polishing it before I make any decisions about it. try5 started on peep to polish the chimera. Tue Jun 17 21:05:36 PDT 2008 Kevin Karplus try5-opt3 and and try3-opt3 score very similarly in both undertaker and (after gromacs and repacking) in rosetta. I think I'll do a polishing run from gromacs optimized models. Wed Jun 18 13:51:18 PDT 2008 Kevin Karplus try6 likes try6-opt3, but (somewhat surprisingly), rosetta doesn't particularly care for try6-opt3.gromacs0.repack-nonPC, preferring try5-opt3.gromacs0.repack-nonPC by quite a bit. try6 is very close to try3, so should replace it in our list. I currently favor submitting ReadConformPDB T0451.try5-opt3.gromacs0.repack-nonPC.pdb # < chimera-try3-try4 ReadConformPDB T0451.try6-opt3.pdb # < try3-opt3.gromacs0 ReadConformPDB T0451.try4-opt3.pdb # < T0451.MQAC1-opt3.gromacs0 < Zhang-Server_TS1 ReadConformPDB T0451.MQAU1-opt3.pdb # from SAM-T08-server_TS1 ReadConformPDB T0451.try1-opt3.pdb I'm mull over this for a while, but I think I'm basically done here. The top hit (1s5aA) forms a tetramer according to PQS, but a monomer according to PDB. The dimer of dimers is not very convincing as a multimer though. 1tuhA is a monomer [PQS] 1nwwA is a dimer {PQS] 2bngA is a hexamer (trimer of dimers) [PQS] Thu Jun 19 12:00:24 PDT 2008 SAM-T08-MQAO hand QA T0451 Submitted Thu Jun 19 12:00:24 PDT 2008 SAM-T08-MQAU hand QA T0451 Submitted Thu Jun 19 12:00:24 PDT 2008 SAM-T08-MQAC hand QA T0451 Submitted Tue Jul 1 08:59:27 PDT 2008 Kevin Karplus The biggest differences between our current "best" models seem to be in the phase of the strand at G73-N85. I think that the strand is upside down in MQAC1 and try5-opt3, even though that is currently scoring the best, perhaps because it lacks the bulges of the other models. I think I may try taking G73-S78 from MQAU1, and the rest from try6-opt3, then trying to close gaps on the resulting chimera. Tue Jul 1 09:28:59 PDT 2008 Kevin Karplus try7 attempting to clean up the chimera-try6-MQAU1. Tue Jul 1 10:21:56 PDT 2008 Kevin Karplus try7 was not particularly successful, scoring worse than try2 through try6. The problem is one very bad break before Q75. I think that problem was that clashes were hurting worse than the breaks, so the G74 residue was swung out of the way. Perhaps I need to turn clashes down to close that break. Tue Jul 1 11:45:12 PDT 2008 Kevin Karplus try8 still has problems with clashes around Q75, but the strand looks pretty good. I'm not sure how best to move around T72, G73, I74, Q75. Should I "explode" them, moving them far apart, and let undertaker try to reassemble them? Wed Jul 2 07:55:24 PDT 2008 Kevin Karplus Before trying an exploded model, I'll see if just turning up clashes and breaks fixes try8. try9 started on peep. Wed Jul 2 10:25:34 PDT 2008 Kevin Karplus try9-opt3 has closed the breaks and reduced the clashes substantially, without apparently hurting other costs much---it now scores better than try6 with the try9 costfcn. I think I'm ready to submit. ReadConformPDB T0451.try9-opt3.pdb # < try8-opt3 < try7-opt3 < chimera-try6-MQAU1 chimera-try6-MQAU1 mostly from T0451.try6-opt3.pdb G73-S78 from MQAU1-opt3 ReadConformPDB T0451.try5-opt3.gromacs0.repack-nonPC.pdb # < chimera-try3-try4 # T0451.try3-opt3.pdb # < try2-opt3.gromacs0.repack-nonPC < align(1tuhA+2geyA) ReadConformPDB T0451.try6-opt3.pdb # < try3-opt3.gromacs0 < try2-opt3.gromacs0.repack-nonPC < align(1tuhA+2geyA) ReadConformPDB T0451.try4-opt3.pdb # < T0451.MQAC1-opt3.gromacs0 < Zhang-Server_TS1 ReadConformPDB T0451.MQAU1-opt3.pdb # < SAM-T08-server_TS1 I started writing the following text for the "method" section, when I realized that I might want to make a try4-try9 chimera, with mostly try4, but try9 for T64-T87. Most of the models I'm submitting for T0451 are from SAM and undertaker, not meta-server models. Model 4 is a meta-server model, based mainly on Zhang-Server_TS1, but the rest are all SAM+undertaker models (even model 5, which is a metaserver model, but based mainly on SAM-T08-server_TS1). The main difference between the SAM models and the Zhang-Server model is the bulge at D80, which is not predicted by the Zhang-Server. The bulge is there mainly to get better burial for L79, but it changes the strand phase. In other respects, I think I like the Zhang-Server model better. Model 1 T0451.try9-opt3.pdb # < try8-opt3 < try7-opt3 < chimera-try6-MQAU1 chimera-try6-MQAU1 mostly from T0451.try6-opt3.pdb G73-S78 from MQAU1-opt3 2 T0451.try5-opt3.gromacs0.repack-nonPC.pdb # < chimera-try3-try4 # T0451.try3-opt3.pdb # < try2-opt3.gromacs0.repack-nonPC < align(1tuhA+2geyA) 3 T0451.try6-opt3.pdb # < try3-opt3.gromacs0 < try2-opt3.gromacs0.repack-nonPC < align(1tuhA+2geyA) 4 T0451.try4-opt3.pdb # < T0451.MQAC1-opt3.gromacs0 < Zhang-Server_TS1 5 T0451.MQAU1-opt3.pdb # < SAM-T08-server_TS1 Wed Jul 2 12:23:09 PDT 2008 Kevin Karplus OK, rearrange to put try10 2nd and remove try4. ReadConformPDB T0451.try9-opt3.pdb # < try8-opt3 < try7-opt3 < chimera-try6-MQAU1 # ReadConformPDB T0451.try8-opt3.pdb # < try7-opt3 < chimera-try6-MQAU1 # ReadConformPDB T0451.try7-opt3.pdb # < chimera-try6-MQAU1 ReadConformPDB T0451.try10-opt3.pdb # < chimera-try4-try9 # ReadConformPDB T0451.try4-opt3.pdb # < T0451.MQAC1-opt3.gromacs0 < Zhang-Server_TS1 # ReadConformPDB T0451.MQAC1-opt3.gromacs0.pdb # from Zhang-Server_TS1 ReadConformPDB T0451.try5-opt3.gromacs0.repack-nonPC.pdb # < chimera-try3-try4 ReadConformPDB T0451.try6-opt3.pdb # < try3-opt3.gromacs0 < try2-opt3.gromacs0.repack-nonPC < align(1tuhA+2geyA) # ReadConformPDB T0451.try3-opt3.pdb # < try2-opt3.gromacs0.repack-nonPC < align(1tuhA+2geyA) # ReadConformPDB T0451.try2-opt3.pdb # < align(1tuhA+2geyA) ReadConformPDB T0451.MQAU1-opt3.pdb # from SAM-T08-server_TS1