Kevin Karplus 10 May 1998 t50 is PIR2:I406658 For t50, the target98 method didn't find any really close homologs, but 1bmtA came in on the fourth iteration. Component S (t50) is supposed to have high similarity to B12-binding domains of methylmalonyl CoA mutase (1reqA) and methinonine synthase (1bmtA). Very high scorers with t98 are (1reqA, 1reqC, 2reqA, 2reqC, 3reqA) => fssp 1reqA (1bmtA, 1bmtB) => fssp 1bmtA 1reqA and 1bmtA are weakly related (Z-score 3.6). Stronger relations to 1reqA include 1reqB, 1ao4A, 1ceo, 1tmy, 1nal13, 1burA, 1dapA, 1xyzA, 8abp, 1aq0A, 1ntr, 1amk, 1iibA, 1lucA, ... none of which score at all well. With the template models, 1bmtA_2 scores excellently (-75.77) though 1bmtA as a whole chain does not. Summing in both directions gives 8 significant hits: t50 1reqA -99.310 t50 1reqC -99.310 => 1reqA t50 2reqA -99.310 => 1reqA t50 2reqC -99.310 => 1reqA t50 3reqA -99.310 => 1reqA t50 1bmtA -94.170 t50 1bmtB -94.170 => 1bmtA t50 1bmtA_2 -75.770 We have two competing structures for t50, both of which seem excellent matches. We will certainly need fssp alignments to get decent alignments to either. 1bmta_2 is SCOP class 3.13.5.1.1 From karplus@cse.ucsc.edu Mon May 11 01:25:00 1998 Return-Path: karplus@cse.ucsc.edu Date: Mon, 11 May 1998 01:24:59 -0700 From: Kevin Karplus To: markd@cse.ucsc.edu Cc: karplus@cse.ucsc.edu Subject: still more alignments to create 1bmtA and 1reqA for t50. -------------------------------------------------- Kevin Karplus 11 May 1998 Although 1bmtA and 1reqA have different structures, there is a part of 1reqA that looks like the beta-sheet braced by alpha helices that is the matching part of 1bmtA_2. If this is the part of 1reqA that alignment is to, then we may have agreement. 1bmtA_2 is "741-896:A" in RASMOL. The interesting corresponding region in 1reqA is 561-728:A The 1bmtA alignment 1bmtA-t50.pw.a2m looks very good in 3d--I just have to move the 5 residues at the end back onto the helix. The two insertions appear on adjacent surface loops. Sure enough, the 1reqA homology is to the correct part of that sequence, and the t50-1reqA alignment looks pretty good, with only one insertion in a helix that needs to be moved to the end of the helix (but which end?) I've just built a target98 alignment for 1reqA, to try aligning the other way. There are several somewhat different alignments to choose between for this target, though the differences are fairly small. -------------------------------------------------- 24 May 1998 Kevin Karplus With one little glitch the alignment 1reqA-t50-fssp-global looks good. The alignment is easily fixed by aligning the 3 mid-helix residues that are unaligned. 1reqA-t50-const-global looks even better (no fixes needed). 1bmtA-t50-const-global looks good, 1bmta-t50-fssp-global does not. 21 July 1998 Kevin Karplus Moved everything except t50-t98.rdb to "old" and remade with a newer version of the Makefile. Single-blast gets 1reqA=1reqC=2reqA=2reqC=3reqA as top hit, and 1bmtA=1bmtB as second hit. Nothing else relevant. Double-blast gets the same ordering, but gets a moderately strong hit for 1iv[pq][AB]=[3456]upjA=2hp[ef][AB] as well (fssp rep 1fmb). Using the new t98_5 rather than the old t98_6 makes no real difference in the ordering. The third hit from double-blast is not found at all. The 1reqA-t50-const alignment may be a bit shorter than the old one, so the alignments will have to be examined again to see which one to use. I looked at the PDB files for 1req, 2req, and 3req. The best resolution is for 1req (2.0 Ang), but 2req is without the substrate, which may make it more similar to t50, if the substrate interacts at all with the relevant domain. The 1bmt file is only resolved to 3.0 Ang, so 1req still seems to be the best choice. For looking at the tree, the interesting PDB sequences correspond to the following swissprot ids: 1bmtA=METH_ECOLI, 1reqA=MUTB_PROFR. Interestingly, the 1bmtA sequence is in the same subtree as T0050, but 1reqA is in another subtree. Here is the alignment used for the tree: 3D:1BMTA_94:216 T0050_1:137 SW:MUTB_PROFR_594:723 skeqgKTNGKMVIATVKGDVHDIGKNIVGVVLQCNNYEIVDLGVMVPAEKILRTAKEVNADLIGL V D H G I V GV P E A E AD I L .....MEKKTIVLGVIGSDCHAVGNKILDHAFTNAGFNVVNIGVLSPQEVFIKAAIETKADAILL I L G D H G K A GF V E A E feqaeGRRPRILLAKMGQDGHDRGQKVIATAYADLGFDVDVGPLFQTPEETARQAVEADVHVVGV ..S...GLITPSLDEMVNVAKEMERQG.F...T...-.IPLLI..GGATTS..KAHTAVKIEQNY S L G I L GG E ..S...SLYGQGEIDCKGLRQKCDEAG.L...E...G.ILLYV..GGNIVV..GKQHWPDVEKRF S SL G LR D G IL V GG I ..S...SLAGGHLTLVPALRKELDKLG.R...P...D.ILITV..GGVIP-..-----EQDFDEL -.-.....--------...--SG.PT.VY.....V..QNASRTVGV-vaall. P V K.D.....MGYDRVYA...PGTP.PE.VG.....I..ADLKKDLNIE...... G Y PGT L K L R.K.....DGAVEIYT...PGTV.IP.ES.....A..ISLVKKLRA-slda.. PE I KK iPE.SA.....I..SLVKKLRA There may be a slight misalignment at the end of 1reqA (the PE and KK could be conserved, getting 2 more identities). There are some parts near the beginning that match 1bmtA better, but if we go with a single template, 1reqA looks more promising. (If we were doing 3D refinement, we might make constraints for both templates.) Fri Jul 24 17:20:44 PDT 1998 Maybe try adding last 25-30 residues to beginning of t51 to improve chances of prediction there. 26 July 1998 Since 1reqB now comes up as strongest hit for t51, I lean strongly toward 1reqA for t50. 28 July 1998 Unfortunately, the part of 1reqA that we match does NOT interact with 1reqB, except through a long ligand (coenzyme B12). My best guess at a 1reqA alignment is t50-1reqA-hand.a2m. Fri Aug 7 17:12:50 PDT 1998 Redid the hand alignment, getting 1reqA-t50-hand1.a2m. I compared it with the old one, and they agree except for slight differences in what to unalign. I think I lke the 1reqA-t50-hand1.a2m alignment infintesimally better. 1bmtA-t50-hand.a2m 31 conserved and 2 1-residue inserts 1reqA-t50-hand1.a2m 41 conserved a 2-vs-7 unalignment, a 1 residue insert and a 1 residue gap I like the 1reqA alignment better, even with the extra gaps. From venc@september.llnl.gov Wed Aug 26 11:53:28 1998 Return-Path: venc@september.llnl.gov Date: Wed, 26 Aug 1998 11:52:29 -0700 From: venc@september.llnl.gov (Ceslovas Venclovas) To: karplus@cse.ucsc.edu Subject: Update for T0050 Dear Predictor, Today (August 26) PDB released coordinates of Glutamate Mutase (B12-Binding Subunit) - NMR, Minimized Average Structure, accession code 1BE1. This protein is closely related to the target T0050. To let anyone interested make use of this newly available structural information the expiration date for T0050 has been moved to September 15. Sincerely, Ceslovas Venclovas for CASP3 organizers -- Protein Structure Prediction Center, Lawrence Livermore National Laboratory, Livermore, CA 94550 E-mail: venclovas1@llnl.gov Phone: (925) 422-3097 Fax: (925) 423-3608 From venc@september.llnl.gov Mon Aug 24 12:06:57 1998 Return-Path: venc@september.llnl.gov Date: Mon, 24 Aug 1998 12:05:59 -0700 From: venc@september.llnl.gov (Ceslovas Venclovas) To: karplus@cse.ucsc.edu Subject: Update for T0050 Dear Predictor, It has been brought to our attention that the structure of the protein closely related to target T0050 (82% sequence similarity, no indels) has just been published in the last issue of STRUCTURE: **** How a protein prepares for B12 binding: structure and dynamics of the B12-binding subunit of glutamate mutase from Clostridium tetanomorphum Martin Tollinger, Robert Konrat, Brent H Hilbert, E Neil G Marsh and Bernhard Kraeutler Structure 6 (1998), 1021-1033. (15 August 1998) **** This issue is also expected to appear soon on the Web at URL: http://biomednet.com/library/jstr REMINDER! Please remember that before target expiration date you can supersede any of your submitted models for that target without restrictions. NEW! You can easily check the list of currently accepted models from your group using viewer on CASP3 web page: http://PredictionCenter.llnl.gov/casp3/models/casp3-models.html Sincerely, Ceslovas Venclovas for CASP3 organizers -- Protein Structure Prediction Center, Lawrence Livermore National Laboratory, Livermore, CA 94550 E-mail: venclovas1@llnl.gov Phone: (925) 422-3097 Fax: (925) 423-3608 Fri Aug 28 12:24:02 PDT 1998 The 1be1 alignment scores only a tiny bit better than the 1reqA ones: 1be1/t50-1be1-global 1be1 137 -109.12 -111.26 1reqA/t50-1reqA-global 1reqA 727 -109.59 -106.54 1reqA/t50-1reqA-vit 1reqA 727 -104.31 -103.11 1be1/t50-1be1-vit 1be1 137 -104.27 -101.76 But the number of conserved residues goes up enormously (only 24 residues different in the 137-long gapless alignment), so we should probably resubmit this.