Mon Jun 2 09:25:32 PDT 2008 T0427 Make started Mon Jun 2 09:27:27 PDT 2008 Running on peep.cse.ucsc.edu Make started Mon Jun 2 09:28:18 PDT 2008 Running on cheep.cse.ucsc.edu Mon Jun 2 12:22:07 PDT 2008 Kevin Karplus pdb_blast gets weak hits to 1hr7B and 1hr6B (0.017) The HMMs also find these (d.185.1.1) domains, but more confidently (E-value 7e-60). Tue Jun 3 09:28:18 PDT 2008 Kevin Karplus The top alignments and the try1-opt3 model are all in close agreement. There are, however, some horrendous breaks: T0427.try1-opt3.pdb.gz breaks before (T0427)L110 with cost 27.7204 T0427.try1-opt3.pdb.gz breaks before (T0427)F85 with cost 15.0672 T0427.try1-opt3.pdb.gz breaks before (T0427)A126 with cost 14.1088 T0427.try1-opt3.pdb.gz breaks before (T0427)G310 with cost 9.51138 T0427.try1-opt3.pdb.gz breaks before (T0427)T38 with cost 8.54904 T0427.try1-opt3.pdb.gz breaks before (T0427)A411 with cost 5.23427 Upping the breaks weight and rerunning from alignments is probably worth doing. Tue Jun 3 16:32:08 PDT 2008 Kevin Karplus The try2 run on peep is taking forever (in SCWRL). I should be sure not to run so many SCWRL on subsequent runs!! Tue Jun 3 17:19:00 PDT 2008 Kevin Karplus In 7 hours, the try2 run has only gotten through about 73 alignments, 69 of which timed-out in scwrl, which is not even two of the templates! I think I need to kill try2 and run again without scwrling all the templates. Tue Jun 3 17:23:42 PDT 2008 Kevin Karplus try2 run killed, try2.under edited to change -scwrl scripts to -noscwrl scripts, and started again on peep. Tue Jun 3 19:20:49 PDT 2008 Kevin Karplus typo introduced in try2.under, job ended earlier. Trying again on peep. Tue Jun 3 21:48:33 PDT 2008 Kevin Karplus Yet another typo. Trying again on peep. Wed Jun 4 08:36:04 PDT 2008 Kevin Karplus The best-scoring model with try2.costfcn is try2-opt3.gromacs0.repack-nonPC. Breaks are still bad. I woner if I should try doing the two domains separately? This could cause problems with the packing of the helices K147-F160 and A365-M378, since these helices have a buried face that packs against the other sheet. Maybe I should just do a polishing run with breaks and clashes turned up. Wed Jun 4 08:44:10 PDT 2008 Kevin Karplus try3 started on the moai cluster, with try2.sheets and try2.helices as part of the cost function. Wed Jun 4 18:03:32 PDT 2008 Kevin Karplus try3-opt3.gromacs0 is the best-scoring with the try3 costfcn, but it still has bad breaks. Sun Jun 8 11:06:30 PDT 2008 SAM-T08-MQAO hand QA T0427 Submitted Sun Jun 8 11:06:30 PDT 2008 SAM-T08-MQAU hand QA T0427 Submitted Sun Jun 8 11:06:30 PDT 2008 SAM-T08-MQAC hand QA T0427 Submitted Mon Jun 9 21:01:28 PDT 2008 Kevin Karplus I ran the metaservers with the default try1 costfcn (I may want to rerun with try3.costfcn). MQAC1-opt3 seems to be based mainly on MUProt_TS1 MQAU1-opt3 seems to be based mainly on FALCON_TS3. try3 likes best MQAU1-opt3.gromacs0 MQAC1-opt3.gromacs0.repack-nonPC try3-opt3.gromacs0 The models are all fairly similar. I'll do another pair of metaserver runs with try3 instead of try1 as the costfcn. Sat Jun 14 14:55:01 PDT 2008 Kevin Karplus FOO! the meta-server runs for MQAU3 and MQAC3 still said MQAU1 and MQAC1 in the scripts---so got misnamed and so did not run through gromacs and rosetta. I fixed the names and am rerunning the gromacs and rosetta tweaks. Sat Jun 14 15:04:24 PDT 2008 Kevin Karplus Even without gromacs and rosetta, MQAU3-opt3 is scoring best with the try3 costfcn. MQAC3 is based on MuProt_TS1 MQAU3 is based on FALCON_TS3 (the same models that the MQAC1 and MQAU1 optimizations favored). Sat Jun 14 15:08:58 PDT 2008 Kevin Karplus MQAU3-opt3.gromacs0 is now the favored model with try3.costfcn The MQAU3 and MQAC3 metaserver runs look better than try3, so I think I should now do a polishing run from all models and pick out some to submit. Sat Jun 14 20:07:37 PDT 2008 Kevin Karplus try4-opt3, optimized from MQAU3-opt3.gromacs0, has the new best score, though rosetta still prefers T0427.MQAC3-opt3.gromacs0.repack-nonPC Sat Jun 14 20:28:43 PDT 2008 Kevin Karplus I don't think I'll make further progress, so I'll submit the following models: 1 T0427.try4-opt3.pdb