Tue Jul 11 09:13:08 PDT 2006 T0371 Make started Tue Jul 11 09:14:03 PDT 2006 Running on lopez.cse.ucsc.edu Tue Jul 11 09:18:11 PDT 2006 Kevin Karplus BLAST gets some excellent full-length hits in PDB: 1zjjA 28% over 250 residues 1.2e-17 1pw5A 27% over 260 residues 1.8e-13 1vjrA 27% over 260 residues 2.3e-13 2c4nA 26% over 254 residues 2.8e-11 ... Wed Jul 19 10:52:42 PDT 2006 Zack Sanborn Well, this target has a hard deadline on August 1st, so I'd better start taking a look at it. From all.breaks.gz, we see that try1-opt2 has several large breaks (as expected from the automatic try): T0371.try1-opt2.pdb.gz breaks before (T0371)E226 with cost 5.87028 T0371.try1-opt2.pdb.gz breaks before (T0371)I181 with cost 3.15494 T0371.try1-opt2.pdb.gz breaks before (T0371)A180 with cost 2.99643 T0371.try1-opt2.pdb.gz breaks before (T0371)L156 with cost 2.91304 T0371.try1-opt2.pdb.gz breaks before (T0371)I74 with cost 1.95408 T0371.try1-opt2.pdb.gz breaks before (T0371)T171 with cost 0.889432 T0371.try1-opt2.pdb.gz breaks before (T0371)L24 with cost 0.874031 T0371.try1-opt2.pdb.gz breaks before (T0371)P11 with cost 0.835823 T0371.try1-opt2.pdb.gz breaks before (T0371)Y27 with cost 0.749308 T0371.try1-opt2.pdb.gz breaks before (T0371)R159 with cost 0.640562 (Remaining 25 breaks are less than 0.5). So, we should definitely try to heal those breaks. Usually, this is done by taking a GROMACS optimized structure and polishing it. This might be a good idea as the GROMACS model (try1-opt2.gromacs0.pdb) has only 17 breaks and its highest is 3.35177. Looking at the structure, I see a tiny beta-hairpin that flies off into nowhere (residues Asp178 - Pro173). We might need to tighten it down, but it is predicted to be coil. I think if I increase the dry5 weights, Undertaker will attempt to pack that coil a little better. I'm going to start a pair of trys: try2 will optimize try1-opt2, increasing the penalties for breaks, soft clashes and dry5 weights. try3 will optimize try1-opt2.gromacs0.pdb, increasing the penalties and weights identically as try2. Wed Jul 19 17:24:19 PDT 2006 Zack Sanborn try2 and try3 have finished. try2-opt2 is now the highest scoring by a little bit using the unconstrained cost function. No try3 model was liked by Undertaker using this cost function. Both models still have significant breaks which I would really like to reduce if possible. Also, I notice that model 1 from undertaker-align.pdb had a nice beta sheet from residues D169-A182. I think I will make a chimera of try2-opt2 and the residues D169-A182 from this model. I like this sheet better than what is in the try2-opt2 model. I've made a chimera as described, copying-and-pasting residues D169-A182 (those containing the nice beta hairpin) from model 1 of undertaker-align.pdb (from alignment to ....) into our best-scoring model try2-opt2. The chimera can be found at decoys/T0371.chimera.pdb.gz. I've started an optimization for the chimera, try4. I've upped the penalties for breaks and soft_clashes on camano. I've also started an optimization from all models (by including read-pdb.under) as try5 on vashon. Wed Jul 19 18:23:00 PDT 2006 Zack Sanborn Okay, I don't know why I'm getting the following error for try4: # iteration 0 filling pool to have 20 conformations undertaker: # ConformationPool.h:49: Conformation*& # ConformationPool::conform(int): Assertion `0<=sub && sub