Mon Aug 17 09:52:28 PDT 1998 There is some reason to doubt that t82 is "homologous sequence of known structure"---at least, sequence-level homology is not obvious. We may want to take this as a hint that "new fold" is wrong. wu-blast does not find an obvious homology (best scores 1maeL=1mafL-2madL -1.171183) double-blast does not find an homologous sequences t82.t98_6 does not include any PDB chains. The top hits with t82.t98_6 are not very impressive: 1beb[AB] -5.240 1bebA 1jsg -4.530 1jsg 1mh1 -4.270 1mh1 2pil -4.220 2pii? 1ay2 -3.990 1ay2 1ece[AB] -3.990 1eceA 2sas -3.640 2sas 1frj -3.600 6fd1 1ne[qr] -3.600 1neq These seem to be all over the place--no clear consensus. Using target98 library, no good hits: 2bbkL -4.280 1aie -4.160 1hstA -3.810 1olgA -3.480 1chl -3.420 2spcA -3.270 1bi6H_1 -3.110 Summing both ways, still no great ones: 1aie -5.79 1aie fragment of P53 1eceA -5.51 1eceA 1olgA -5.42 1olgA P53 1beb[AB] -5.33 1bebA 1neq -4.7 1neq DNA-binding 1afrA -4.57 1afrA Looking at the fssp library: 1fow -5.520 1aoo -4.940 1phd -4.490 2bbkL -4.320 1fleI -4.080 1at0 -3.900 1pnbB -3.480 Looking at the target98-mixed library: 2bbkL -4.200 1aie -4.170 1olgA -3.810 1amp -2.960 The repeated occurrence of 1aie and 1olgA (two different domains of P53) is rather interesting. Here are the best non-self scores for the alignments that looked promising: 1bebA/t82-1bebA-vit 1bebA 156 -8.22 -8.71 Matches a beta-hairpin and 1 strand of next sheet 1neq/t82-1neq-vit 1neq 74 -4.71 -5.87 11-residue match 1olgA/1olgA-t82-global T0082 190 -6.27 -5.75 27-residue match, with 8-residue gap 1olgA/1olgA-t82-post T0082 190 -6.27 -5.75 1olgA/1olgA-t82-vit T0082 190 -3.31 -5.59 1aie/1aie-t82-global T0082 190 -6.80 -5.57 19-residue match (1 helix) 1aie/1aie-t82-post T0082 190 -6.80 -5.57 1aie/1aie-t82-vit T0082 190 -4.68 -5.43 1eceA/t82-1eceA-vit 1eceA 358 -7.42 -5.20 28-residue match (5 conserved) dimerization site 1bebA/t82-1bebA-global 1bebA 156 -2.90 -5.06 approx 70-residue match, scattered 1bebA/t82-1bebA-post 1bebA 156 -2.90 -5.06 1aie/t82-1aie-vit 1aie 31 -4.33 -4.64 1olgA/t82-1olgA-vit 1olgA 42 -4.03 -4.64 2bbkL/2bbkL-t82-fssp-global T0082 190 -4.16 -4.49 looks pretty good, for first 120 residues 2bbkL/2bbkL-t82-global T0082 190 -4.09 -4.31 2bbkL/2bbkL-t82-post T0082 190 -4.09 -4.31 2bbkL/2bbkL-t82-vit T0082 190 -5.98 -4.21 1neq/1neq-t82-fssp-global T0082 190 -3.30 -4.11 1bebA/1bebA-t82-fssp-global T0082 190 0.72 -3.51 1neq/1neq-t82-fsspt98-global T0082 190 -2.88 -3.38 1neq/1neq-t82-const-global T0082 190 -2.97 -2.78 The 2bbkL-t82-fssp-global alignment looks pretty good, but has several predicted helices as beta strands. I tweaked it to get 2bbkL-t82-hand1.a2m. Still, this seems like an unlikely structure for a ribonuclease. PRODOM gives only one domain for this protein, shared with many other ribonucleases. The active site residues are 34, 85, 89 (by similarity to something unspecified). 34: ftiHglw 85,89: fwshEwtkHgtccse The ribonucleases of known structure are in "known ribonucleases" Their fssp reps are 1eriA 1rgeA 1rthA 2rn2 3dni 7rsa 9rnt They all seem to have an exposed helix over a beta sheet, though there is considerable variation in what is behind the beta sheet. I can hand-edit alignments to 2rn2 (t82-2rn2-hand1.a2m) and 1eriA (t82-1eriA-hand1.a2m) to get ok residue conservation, but the active-site residues are not clustered. It may be worth trying again with the remote_4 alignment, since that seems to have a few more members of the ribonuclease family, but not a lot of crap. 26 August 1998 Christian Under additional information on the CASP t82 web page, it states "RNase Rh, of which coordinates is in the PDD (1BOL), has sequence similarity with RNase MC1" 1bol is definitely the structure. Here are the non-self scores: 1bol-t82-global.pw.dist:T0082 190 -118.24 -116.18 1bol-t82-post.pw.dist:T0082 190 -118.24 -116.18 1bol-t82-vit.pw.dist:T0082 190 -113.85 -111.73 t82-1bol-global.pw.dist:1bol 222 -90.82 -91.20 t82-1bol-post.pw.dist:1bol 222 -90.82 -91.20 t82-1bol-vit.pw.dist:1bol 222 -77.85 -76.35 From cbarrett@cse.ucsc.edu Wed Aug 26 14:42:21 1998 Mail-from: From cbarrett@cse.ucsc.edu Wed Aug 26 14:42:21 1998 Return-Path: cbarrett@cse.ucsc.edu Received: from moo.cse.ucsc.edu (moo.cse.ucsc.edu [128.114.134.35]) by services.cse.ucsc.edu (8.8.4/8.6.12) with SMTP id OAA29481 for ; Wed, 26 Aug 1998 14:42:21 -0700 (PDT) Received: from localhost (cbarrett@localhost) by moo.cse.ucsc.edu (8.6.10/8.6.12) with SMTP id OAA14083 for ; Wed, 26 Aug 1998 14:42:21 -0700 X-Authentication-Warning: moo.cse.ucsc.edu: cbarrett owned process doing -bs Date: Wed, 26 Aug 1998 14:42:20 -0700 (PDT) From: Christian Barrett X-Sender: cbarrett@moo To: karplus@cse.ucsc.edu Subject: odd behavior Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Check out t82. The CASP web page, under "Additional info", actually gives us a homologous structure --- 1bol. It wasn't in our library so I built it and, sure enough, the alignments are in the -100 range. But what is odd is that our methods didn't pick this structure up until t82.remote_4-varh50-pdb.rdb:t82 1bol -77.720 The alignment shows a couple of very conserved regions that should have generated some sort of signal. The blast methods also fail to find this structure. Christian From karplus@cse.ucsc.edu Wed Aug 26 17:33:27 1998 Return-Path: karplus@cse.ucsc.edu Date: Wed, 26 Aug 1998 17:33:26 -0700 From: Kevin Karplus To: cbarrett@cse.ucsc.edu Cc: karplus@cse.ucsc.edu In-reply-to: (message from Christian Barrett on Wed, 26 Aug 1998 14:42:20 -0700 (PDT)) Subject: Re: odd behavior I think that the reason we didn't find 1bol is that it was only released 5 Aug 1998, and wasn't in the database when we ran the tests, It looks like the database was last updated 11:52 Aug 24, so the previous version was downloaded around noon on Aug 17, and indeed that is the interval between when the blast and t98_6 runs and the remote_4 runs was made. I'm going to redo the target98 and remote_4 runs, so that 1bol will be properly found. Wed Aug 26 17:37:34 PDT 1998 Kevin Karplus Now wu-blast finds 1bol with no problems: T0082 1bol -30.495725 double-blast also finds 1bol and nothing else: T0082 1bol -50.4746 3.1e-34 1.2e-22 gi|532754_31:228 RNRH_RHINI (which seems to be 1bol) comes in already on t82.t98_2.a2m, as it did before. t82.t98_6 now finds 1bol with score -82.19 Summing both ways doesn't change anything, since it isn't in our set of library models yet. The 1bol-t82-global model looks pretty good, and I tweaked it a bit to get 1bol-t82-hand1. I like this alignment and am willing to submit it.