18 June 1998 Kevin Karplus There are several high-scoring chains, including chain score fssp rep 1adr -101.2 1r69 1r69 -92.51 1r69 1perL -47.780 1r69 1perR -47.780 1r69 1pra -47.780 1r69 1rpeL -47.780 1r69 1rpeR -47.780 1r69 2or1L -47.780 1r69 2or1R -47.780 1r69 1r63 -47.550 1r69 1lliA -45.64 1lmb3 2r63 -45.190 1r69 1zug -44.210 1r69 2cro -44.210 1r69 3croL -44.210 1r69 3croR -44.210 1r69 1lliB -33.160 1lmb3 1lmb3 -30.670 1lmb3 1lmb4 -30.670 1lmb3 1lrp -30.670 1lmb3 The two hits (1r69 and 1lmb3) are similar structures (Z-score 8.0 in FSSP). We basically have a choice of any of the 1r69 homologs, with 1adr initially looking the most promising. I don't see how the submitters can claim "no known homology". One minor problem: T64 and T65 are a heterodimer, and the obvious homologs of T64 are homodimers or monomers. -------------------------------------------------- 19 June 1998 The SCOP family is "Phage repressors" and contains lambda C1 repressor, DNA-binding domain 1lmb[34] 1lrp 1lli[AB] 434 C1 repressor, DNA-binding domain 1r69 1per[LR] 2orl[LR] 1rpe[LR] 1pra 1r63 2r63 cro 434 2cro 3cro[LR] 1zug p22 C2 repressor, DNA-binding domain 1adr cro lambda repressor 4cro[ABCDEF] 1cro[OACD] 1cop[DE] 1orc NER 1neq 1ner The superfamily ""lambda repressor-like DNA-binding domains" also contains Oct-1 POU-specific domain 1oct 1pou (note: fssp rep is 1au7A) bacterial repressors 1bdi 1pnr 1pru 1prv 1bdh 1lccA 1lcdA 1lbg 1lqc 1uxd 1uxc The t64.t98_6 alignment picks up 1adr, 1r63, 2r63, 2cro, 3croL, 3croR. (These are first picked up in iteration 5.) The t64.remote_4 alignment picks up these plus 2mysc1, 1uxc, 1uxd, 1rh2A2, 1rh2C3, 1rh2B3, 1lccA, and 1lqc. The 2mysC1 is an EF-hand, a completely different fold. In the evolutionary tree built from t64.t98_6, 1adr is the closest structure to the target sequence. -------------------------------------------------- 19 June 1998 Kevin Karplus Moved the target98 alignments to old-sam, and created new ones using the latest version of SAM. Also, dropped the creation of the remote alignments and used the t64.t98_5 alignment as the master alignment, rather than t64.t98_6, since it already contains several PDB sequences. The scores are slightly different with the new sam, but the same sequences come out on top. The alignment with 1ADR matches the first 66 bases of t64 excellently, but none of the repressors is long enough to cover the rest. The longest is 1lmb3 (87 bases), but it might be worth looking at 1au7A, which has a moderately good alignment with 1r69 (Z=6.1) and is 130 long. The 1au7A sequence and model do not score super well, but 1au7A is in the top 60 of t64-sum.rdb and is in the same superfamily as the higher-scoring sequences. The t64-1lmb3.pw alignment doesn't look very promising---it isn't as clean as the 1adr alignments, and the extra helix on the end is not a good match to the two helices of t64. The t64-1au7A.pw alignment is similarly disappointing, though the extra helices there could perhaps be related to the extra helices in t64. 16 July 1998 Kevin Karplus single-blast gets strong hits for 2or1[RL], 1per[RL] 1pra, 1r63, 1r69, 1rpe[RL] (all -10.03, all identical to 1r69). 1adr does ok, but not quite as well (-4.906) double-blast finds the same set (-12.226), then 3cro[RL],1zug,2cro,2r63 (-11.64) then 1adr (-11.33) rescoring using t64.t98_4 23 July 1998 1adr still scores best, then 1r69 close behind, then a big gap to 1pre[LR]=1pra=1rpe[LR]..., probably because of 1r69 being selected as the template for that group of sequences. 1adr scores just 0.13 behind 1r69 with the t64.t98_4 model, so 1adr and 1r69 are really neck-in-neck. The best score with a template model is for 1adr-t64-global. There is no constrained model for 1adr, since it is in the 1r69 FSSP file, with 33% residue identity. The constrained alignment for 1r69 has not been built yet, so I'll try that now---hmm, no improvement. In fact it scores worse than the 1r69-t64-global alignment. Since 1ADR is just an NMR of the protein in solution (not an X-ray crystal), I wonder whether using the 1R69 pdb file (a 2-Angstrom X-ray structure) is better. There are 2 1-residue insertions relative to 1r69 and 46 unaligned residues off the end, while there is only 1 1-residue insertion realtive to 1adr, and only 38 unaligned residues off the end. I'll also try searching with just the C-terminus of t64 (t64-end) to see if we can come up with any matches for that part. Single-blast scored 1slm highest (though -1 is a VERY weak score). Double-blast found nothing. Top for the t64-end target model is 1tnc at -4.07, then 1tcf=1tn4=2tn4 at -3.97. 1tnc is a theoretical model for troponin c; 1tn4 is troponin c and has 1ncx as its fssp representative. Question---does t64 bind calcium? The troponins did not score well on the whole t64 model, but t64-end.remote_4 brings in the whole set of ef-hands. Top with the template models is 1tpt at -3.30 (thymidine phosphorylase, fssp rep 1tpt). 1tpt actually scored modestly on the whole chain (-2.78), so might be worth looking at. Summing does not raise these scores. Blast's find, 1SLM (stromelysin-1 fragment), scores near the top, with -3.17, but only gets -1.57 for the full model. Fri Jul 24 14:49:59 PDT 1998 The top constrained alignments are 1lmb3/1lmb3-t64-const-global.pw.dist:T0064 111 -44.13 -45.24 1r69/1r69-t64-const-global.pw.dist:T0064 111 -22.38 -21.89 1au7A/1au7A-t64-const-global.pw.dist:T0064 111 -4.42 -0.43 (1adr can't have a constrained model, since it isn't an fssp rep.) The best global template models are 1adr/1adr-t64-global.pw.dist:T0064 111 -59.17 -59.89 1r69/1r69-t64-global.pw.dist:T0064 111 -54.14 -56.14 1lmb3/1lmb3-t64-global.pw.dist:T0064 111 -6.56 -5.97 1au7A/1au7A-t64-global.pw.dist:T0064 111 -0.75 2.30 The 1lmb3/1lmb3-t64-const-global alignment looks pretty good, with one 4-residue deletion and 1 1-residue insertion, both in exposed loops. 1lmb3 is bound to DNA, and there is no mention of DNA in the t64+t65 crystal, so it is probalby in an unbound form. The 1adr/1adr-t64-global alignment looks even better, with just 1 1-residue insertion. The 1r69/1r69-t64-global.pw alignment has 2 1-residue insertions and doesn't have as much at the C-terminus. The C-terminus is unconserved random coil in 1adr, so there isn't much point to aligning to it anyway. The 1lmb3 alignment goes further than 1adr, continuing the helix out to YDGQLDSE. This is consistent with the t66 prediction of a helix, [OOPS--these regions don't overlap] so the expectation is that the helix continues on away from the DNA-binding site. I think I'll go with the 1adr-t64-global alignment out to VHTLLD, then switch to the t66 alignment to 1ois. 28 July 1998 Kevin Karplus Based on the t66 alignments, I'll only take the alignment out to the VSHTLL, and then switch to 1ois, taking the alignments from t66. With the hand alignment to 1adr having 2 insertions now, it isn't clear whether 1adr or 1r69 is the better choice!