Thu Jul 29 15:53:40 PDT 2004 T0256 Due 24 Aug 2004 Fri Jul 30 05:33:38 PDT 2004 Kevin Karplus The try1-opt2 model looks pretty good, but it has a large exposed hydrophobic patch. Are the templates dimers? Y99, H5, H62, E109 may be an active site. Sat Jul 31 04:09:35 PDT 2004 Kevin Karplus The side chains could be packed a bit tighter. I wonder whether F85-P90 should be connected as an edge strand to the sheet. Templates from t2k: 1lkdA 1nkiA 1lgtA 1lqpA 1han 1kw3B 1dhy 1gdgA 1f9zA 1f1uA 1f1xA 1mpyA 1fa8A 1qipA 1r9cA 1bh5A 1npbA 1kllA 1ecsA 1ewjA 1jc4A 1kwbB 1q0cA 1qtoA 1bylA 1kmzA 1cjxA 1mh6A Templates from t04: 1fa8A 1f9zA 1qipA 1nkiA 1lqpA 1han 1lkdA 1f1uA 1f1xA 1ecsA 1r9cA 1kw3B 1jc4A 1kllA 1bylA 1qtoA 1cjxA 1k4nA Note: the top hit, 1lkdA, has TWO copies of the domain, so we should probably handle T0256 as a dimer. To get alignments for the dimer, I have requested a VAST structural alignment for try1-opt2: job VS60322 password T0256try1 Tue Aug 3 04:21:24 PDT 2004 Kevin Karplus The top vast hit is to 1han. I've stored the top two alignments to the two domains in 1han/1han-T0256-vast-try1.a2m and 1han/1han-T0256-vast-try1a.a2m. There are a few clashes in dimer-1han-try1.pdb, but not too bad. We should probably do a little polishing to try to pack the sidechains tighter, then do a repack of the dimer with Rosetta. Wed Aug 4 02:17:25 PDT 2004 Kevin Karplus The try1 dimer is looking a bit foamy, but otherwise quite good. There are few conflicts between the monomers, and they fit against each other fairly well. I see no advantage to adding any constraints. Wed Aug 4 05:42:33 PDT 2004 Kevin Karplus Try2 looks fairly good. It scores slightly better than try1, but it may be that the barrel needs to have a slightly larger diameter to pack the back of the sheet (the dimerization interface) a little tighter. The dimer also looks good---it may be time to repack the dimer using rosetta. Sat Aug 14 15:44:23 PDT 2004 Kevin Karplus I repacked the dimer today to create dimer-1han-try2.repack-nonPC.pdb Looking for close contacts in RASMOL: select *:A and within(4.0, *:B) and not hydrogen 146 atoms selected! Looking for bad clashes: select *:A and within(2.0, *:B) and not hydrogen 12 atoms selected! The bad clashes are for M1.O, G2.C, Q67.OE1 P58.CB, K59.CB, T60 (7/14 atoms) Doing the selection the other way around (since repacking does not keep the two chains identical): RasMol> select *:b and within(2.0, *:A) and not hydrogen 12 atoms selected! RasMol> show selected Chain: B Group: MET 1 (C) (2/17) atoms Chain: B Group: GLY 2 (C) (2/7) atoms Chain: B Group: TYR 23 (T) (1/21) atoms Chain: B Group: LYS 59 (T) (1/22) atoms Chain: B Group: THR 60 (C) (5/14) atoms Chain: B Group: GLY 104 (T) (1/7) atoms The conflicts at the Nterminus could probably be resolved by getting one more hydrogen bond in the in initial strand. The current try2-opt2.sheets constraints are SheetConstraint H5 L8 G64 F61 hbond H5 SheetConstraint H5 V10 I46 K51 hbond L6 The strand to extend would be SheetConstraint K4 L8 F65 F61 hbond H5 No, I *don't* like that---it would bury the K4 charges in the interface, with no compensating negative charges. What I suspect happens is that D41 or D43 and K4 attract each other across the interface, though it may be that K4 is attracted more to its OWN D41 and D43. In any case, I don't think I know enough about what is going on in the interface to fix M1-I3 with constraints. The T60-T60 clash may be solvable by extending the strand it is on back one, from SheetConstraint H5 L8 G64 F61 hbond H5 to SheetConstraint H5 T9 G64 T60 hbond H5 Sat Aug 14 21:56:26 PDT 2004 Kevin Karplus try3-opt2 scores better with the try3 costfcn, and with unconstrained.costfn. There are also fewer clashes in the dimer, both before and after repacking with Rosetta. I'll do an unconstrained optimization of try3-opt2, and see if it does any better (there are still some internal clashes that could be removed). Sun Aug 15 08:12:12 PDT 2004 Kevin Karplus try4 scores best with the unconstrained cost funcion, and repacking the dimer with rosetta does not increase the number of clashes over try3. The main remaining clashes are the T60A-T60B ones. dimer-1han-try4.repack-nonPC has the best Rosetta energy of the dimers (they all have terrible scores from the T60 clashes). try3-opt2.repack-nonPC has the best score of the repacked monomers. from-dimer-try4B scores slightly better than from-dimer-try4A with the unconstrained undertaker cost fcn. I'll submit try4-opt2 best undertaker score from-dimer-try4B.pdb Rosetta repacked as dimer try1-opt2 full auto T0256-1lkdA-t2k-local-str2+CB_burial_14_7-0.4+0.4-adpstyle5 best t2k T0256-1f9zA-t04-local-str2+CB_burial_14_7-0.4+0.4-adpstyle5 2nd t04 8th t2k The two different alignments show two different ways for the repeated subdomains to be packed.