Thu Jul 13 11:47:03 PDT 2000 Kevin Karplus >T0112 Ketose Reductase / Sorbitol Dehydrogenase, Bemisia argentifolii tetramer, pH 8.75, known homologs blast finds 1sdg as the best hit---very strong. double-blast finds 1sdg as very strong hit, with other strong hits. Unfortunately, 1sdg is a theoretical model from 1990---it seems like the removal of theoretical models is not working correctly yet. Excluding 1sdg: There are strong blast hits to 1ykf[ABCD] 1bxz[ABCD] (identical sequences) and pretty strong ones to 1ped[ABCD] 1kev[ABCD] (identical sequences) The strongest double-blast hits are to 1ykf[ABCD] 1bxz[ABCD] 1ped[ABCD] 1kev[ABCD] All have 1pedA as the fssp representative, though the 1ykf sequences are only 75% identical (RMSD 0.6 Ang over 351 residues though). 1ped has somewhat better resolution (2.15 Ang vs 2.50 Ang). 1bxz is worse at 2.99, but 1kev has 2.05. We may want to choose as a model the one with the most similar crystallization conditions. 1kev and 1ykf contain NADP, 1ped and 1bxz do not. The t2k alignment includes many PDB sequences (at least 17 distinct ones), and the center of the alignment has moved away from the target, so we may need to use a subfamily to get a more specific alignment. Sat Aug 26 14:47:01 PDT 2000 Kevin Karplus Remade 2track predictions Top hits are now % Sequence ID Length Simple Reverse E-value SCOP 2ohxA 374 -241.90 -214.58 3.9e-90 2.33.1.2.1,3.2.1.1.1 1agnA 373 -241.83 -212.88 2.9e-89 2.33.1.2.2,3.2.1.1.2 1ykfA 352 -200.96 -172.67 6.8e-72 2.33.1.2.4,3.2.1.1.4 1pedA 351 -193.76 -168.14 3.7e-70 2.33.1.2.3,3.2.1.1.3 1qorA 327 -176.44 -150.10 2.5e-62 2.33.1.2.5,3.2.1.1.5 Differ only at subfamily level. We should look at the alignments, but we may want to pull out a subfamily of the target alignment to work with. It looks like 1kevA (=1pedA) may be the closest of the PDB sequences in the tree, but it is hard to tell since the tree-drawing software can't handle so big a tree. The sorted sequences in the target multiple alignment puts 1tehA first of the PDB sequences (fssp rep 2ohxA), but this may just indicate drift in the model building. Note: 1pedA, 1ykfA, and 1agnA are tetramers, but 2ohxA and 1qorA are dimers. Since we have a tetramer, we should probably favor 1pedA, 1ykfA, or 1agnA. Sun Aug 27 12:30:30 PDT 2000 Kevin Karplus I looked at the top-scoring alignments: 2ohxA/T0112-2ohxA-global 2ohxA 374 -265.28 -270.82 1.7e-117 1agnA/T0112-1agnA-global 1agnA 373 -263.82 -266.31 9.0e-116 These looked pretty good, but I moved the gaps around just a little to get better conservation and smaller gaps. My current favorite is 1agnA/T0112-1agnA-global-karplus.a2m, though 2ohxA scored slightly higher. (1agnA has more identical residues after adjusting alignment, and is tetrameric.) The top-scoring alignment with a template model is 1ykfA/1ykfA-T0112-global T0112 352 -239.91 -235.62 1.4e-102 This does not need quite as much editing to get 1ykfA/1ykfA-T0112-global-karplus.a2m, which I think I like better, since it has more identical residues in the core, and does not have to have the two extra beta strands edited out by gap movement that 1agnA does. So now my favorite is 1ykfA/1ykfA-T0112-global-karplus.a2m. 29 Aug 2000 Rachel Karchin Looking at 1ykfA-T0112-global-karplus.a2m and checking areas of disagreement with: T0112-1ykfA-2track-global.pw.a2m using Melissa's program. 2track aligns D in pos. 55 of the target with E in the template -- I edited 1ykfA-T0112-global-karplus.a2m to include this. New version is 1ykfA-T0112-global-karplus-edited.a2m The deletion is in a reasonable position near the middle of a loop in either case. Positions 110-132 of the 2track and global-karplus disagree. global-karplus conserves 4 positions in this range, 2track conserves 2 positions. However chemical properties of residues are better matched on 2track where the only awkward match is L (nonpolar, aliphatic) and T (polar, tiny). the global-karplus has 4 positions with questionable matches: M (nonpolar) and D (polar, charged), F (non-polar, aromatic) and T, E (polar, charged) and A (nonpolar, small), F (aromatic, nonpolar) and R (charged, polar). Other possibilities: 1pedA-T0112-fssp-global.pw.a2m Good conservation but quite not as good as 1ykfA-T0112-global-karplus.a2m. It has a few small badly positioned unaligned regions: on a strand (pos 133-139), on half of a helix (pos 340-345) and on a strand (pos 153-156) 1pedA-T0112-fssp-global.pw.a2m Good conservation. three small unaligned regions that fall on strands. None of these appear much better than 1ykfA-T0112-global-karplus2.a2m