Wed Jun 25 09:11:43 PDT 2008 T0471 Make started Wed Jun 25 09:12:17 PDT 2008 Running on cheep.cse.ucsc.edu Wed Jun 25 18:30:26 PDT 2008 Kevin Karplus Despite a best Evalue of 1.1, this target seems to be a good match to c.66.1.* superfamily. The try1 model is not bad, though some gap closure is needed. Sun Jun 29 10:25:47 PDT 2008 Kevin Karplus The MQAU assessment likes the Zhang-Server models best, then METATASSER_TS5, SAM-T08-server_TS1, pro-sp3-TASSER_TS1, ... The MQAC assessment also likes the Zhang-Server models best, but SAM-T08-server is way down around rank 49 (though the predicted GDT is only down to 50% from 57.6% for Zhang-Server_TS1). I'll run the metaserver predictions to see what models we can come up with. The try2 model still needs a little more polishing, as the try2-opt3.gromacs0 model scores better than try2-opt3, mainly due to fewer clashes. Breaks are still pretty bad. Sun Jun 29 12:56:51 PDT 2008 Kevin Karplus It looks like the moai cluster went down (scheduled power outage) at 12:07, and moai isn't back yet. I'll have to rerun meta_MQAC1 when moai is back online. try3 likes try3-opt3, try2-opt3.gromacs0, try1-opt3.repack-nonPC, MQAU1-opt3.gromacs0.repack-nonPC try1 likes try3-opt3, MQAC1-opt2, try2-opt3, MQAU1-opt3, try1-opt3 Sun Jun 29 16:00:38 PDT 2008 Kevin Karplus MQAU1 optimized SAM-T08-server_TS1 MQAC1 optimized Zhang-Server_TS1 Sun Jun 29 20:17:29 PDT 2008 Kevin Karplus The try1 costfcn prefers MQAC1-opt3.gromacs0, try3-opt3, try2-opt3, MQAU1-opt3, try1-opt3. The try2 costfcn prefers try3-opt3, MQAC1-opt3.gromacs0, try2-opt3.gromacs0, try1-opt3.gromacs0.repack-nonPC, MQAU1-opt3.gromacs0 (Try3 the same as try2, expect for the MQAU1 moedl being MQAU1-opt3.gromacs0.repack-nonPC) It looks like time to try polishing the MQAC1-opt3 model. Sun Jun 29 20:29:23 PDT 2008 Kevin Karplus I started try4 to polish the MQAC1-opt3 model. I included the MQAC1 sheets and the try1 sheets, though the are not quite compatible. The last two strands of the MQAC sheet are more compatible with the secondary structure predictions, so perhaps I should have simply turned off the try1 sheets. I'll do that for try5, when try4 finishes. Sun Jun 29 21:23:24 PDT 2008 Kevin Karplus try4 follows MQAC1, so I expect that try5 will follow try4. I may want to look at each of the edge strands and decide which model to take them from---there may need to be a chimera to get the best edge strand at each end. I'm also wondering if I can curl the tip of the hairpin at R120, to get somewhat better packing, particularly to bury conserved L102. I'm not sure exactly what I want it to do though. Mon Jun 30 10:15:41 PDT 2008 Kevin Karplus Indeed try5 is based on try4-opt3. I don't particularly like the strand around G52-R56 in any of the models, as the loop before it is really too tight for a parallel strand connection. Perhaps this strand should be attached antiparallel??? Tue Jul 1 15:57:27 PDT 2008 SAM-T08-MQAO hand QA T0471 Submitted Tue Jul 1 15:57:27 PDT 2008 SAM-T08-MQAU hand QA T0471 Submitted Tue Jul 1 15:57:27 PDT 2008 SAM-T08-MQAC hand QA T0471 Submitted Sat Jul 12 13:31:35 PDT 2008 Kevin Karplus I wonder if I can shift the final hairpin's relationship to the rest of the sheet in try5, using a relationship more like try3's: try5: SheetConstraint M113 N117 Y126 Y122 hbond K114 1 SheetConstraint A71 S75 A95 I99 hbond S75 1 SheetConstraint L72 S75 V19 V22 hbond Y74 1 SheetConstraint I54 R56 L44 D46 hbond V55 1 SheetConstraint V40 D46 T17 G23 hbond T45 1 try3: SheetConstraint A95 K100 A128 F123 hbond L97 1 SheetConstraint N117 G119 E125 F123 hbond N117 1 SheetConstraint L72 I76 R96 K100 hbond S75 1 SheetConstraint L72 S75 V19 V22 hbond Y74 1 SheetConstraint V43 I47 G53 D57 hbond D46 1 SheetConstraint D41 D46 R18 G23 hbond T45 1 Merged: SheetConstraint V40 D46 T17 G23 hbond T45 1 SheetConstraint V43 I47 G53 D57 hbond D46 1 SheetConstraint L72 S75 V19 V22 hbond Y74 1 SheetConstraint A71 S75 A95 I99 hbond S75 1 SheetConstraint A95 K100 A128 F123 hbond L97 1 SheetConstraint M113 N117 Y126 Y122 hbond E125 1 Sat Jul 12 13:48:25 PDT 2008 Kevin Karplus I'll try the merged sheet constraints for try6, starting with models before try4, so that there may be enough breaks to allow things to move. If that doesn't work, I may try a try3/try5 chimera. Sat Jul 12 18:35:08 PDT 2008 Kevin Karplus The try6 costfcn prefers try5-opt3 and try4-opt3 to try6-opt3, though try6-opt3 gets a slightly better score on the combined sheets. I probably should drop clash and break weights down to move things around. For try7, I'll do the same as for try6, but with more weight on the sheet constraint that has to be formed and lower clash and break weights. Sat Jul 12 20:26:47 PDT 2008 Kevin Karplus try7 has made considerable gains on the combined.sheets score, but it will need gap closure before it is competitive with other models. Actually, I'll try making a chimera with try5, to try to straighten the hairpin, copying the hairpin from try5 into try7. Unfortunately, I haven't got the two models lined up, so there will be terrible breaks and clashes in the chimera. I'll try copying T109-Y124 from try5 into try7, then attempt to close gaps. Sat Jul 12 23:04:05 PDT 2008 Kevin Karplus try8-opt3 does ok, but I need to push it harder to close gaps. try9 will try to close gaps on try8 I think I'm reaching the point of diminishing returns now, and should probably submit this target in the morning. Sun Jul 13 06:50:29 PDT 2008 Kevin Karplus try9 did not succeed in closing the gaps and making the full sheet from try8, so I'll try polishing try7 for try10. Sun Jul 13 08:18:20 PDT 2008 Kevin Karplus I've reached a point of diminishing returns now. I don't have any really great models, and the small differences between the exisitng models are hard for me to resolve. I'll submit ReadConformPDB T0471.try5-opt3.pdb # < try4-opt3 < MQAC1-opt3.gromacs0 < Zhang-Server_TS1 ReadConformPDB T0471.try6-opt3.pdb # < MQAC1-opt3.gromacs0 < Zhang-Server_TS1 ReadConformPDB T0471.try9-opt3.pdb #