23 July 1998 Kevin Karplus Since t66 is just t64 and t65 combined, and we don't have a docking program (and not even a confidant prediction of t65), I don't expect to be able to enter anything for t66. The first part of t64 is clearly DNA-binding (like a repressor), so I've made a chimeric sequence containing the last 39 residues of t64 and all of t65, to see if there is some domain split between the two proteins that gives a docking clue. I don't expect this to pan out, but we don't have the tools to do much else. (I didn't include all of t64, since the repressor match was strong enough to wipe out any other signal.) blast's best shot: 1ois (very weakly: -2.04) double-blast: 1auz=1buz (-4.423) Hmm---these are both fairly new, and look promising, since they are also regulatory factors, The chimeric sequence DOES find homologs in NRP that cross the boundary, so maybe this approach isn't doomed after all. The t66-chim.t98_6 model finds 2ci2I=2sniI -7.3 1ciqA 1exn[AB] -6.46 1exnA 3ci2 -5.29 1ciqA 3ctn -4.56 1ciqA=1cirA -4.49 1ciqA Note: 1auz and 1buz template models may not have been built in time for doing the t66-chim-t98 search with the template models. 24 July 1998 The best results with the template models are for 1ois -8.3 1abv -5.83 2liv -5.8 1ahjB -4.65 1tpt -4.25 1setA_1 -3.64 Summing both ways makes the best 1ois -9.51 1ois DNA topoisomerase i fragment 1exnA -7.32 1exnA 5' exonuclease 2ci2I=2sniI -7.31 1ciqA Chymotrypsin inhibitor 2 2liv -7.19 2liv leucine-isoleucine-valine binding 1exnB -6.46 1exnA = 1sesA -6.03 1sesA seryl-trna synthetase 1abv -5.83 1abv delta subunit f1f0-atp synthase 4hb1 -5.71 4hb1 3ci2 -5.29 1ciqA Chymotrypsin inhibitor 2 1div -5.13 1div ribosomal protein 19 The 1ois alignment has a large gap near where the artificial join ocurred, so I'm trying again with t66-chim20, which adds 20 X residues to make the join. The 1ois-t66-chim-const-global alignment adds a gap of only 7 residues, so I'll try t66-chim7 as well. For t66-chim20, blast finds nothing (well---1slm and 1bod very weakly). Double-blast finds 1auz and 1buz. t66-chim20.t98_6 finds 1ois -6.45 1ois 1jdb[BEHK] -5.46 ? 1pysA -4.45 1pysA phenylalanyl-trna synthetase 1aa2 -3.93 1aa2 beta-spectrin fragment (calponin homology) With the template models, the best matches to t66-chim20 are 2liv -5.68 1eaf -4.72 1tpt -4.54 Summing both ways the best matches to t66-chim20 are 1ois -10.18 2liv -5.68 1jdb[BEHK] -5.46 1aa2 -5.45 For t66-chim7, blast finds 1ois (-6.377)---well into the true-positive region! Double-blast still finds 1auz and 1buz (-4.42285) t66-chim7.t98_6 finds 1ois -9.07 1fos[EG] -4.27 1a02F -4.19 1pysA -3.89 The template models find 1ois t66-chim7 -7.550 1setA_1 t66-chim7 -4.750 1hmf t66-chim7 -4.600 1aab t66-chim7 -4.580 1hryA t66-chim7 -4.420 1tpt t66-chim7 -4.290 Summing both ways moves 1ois up to -16.62. Double-blast didn't find 1ois because of the difference in meaning of E value in PDB and NRP, and because the chimeric sequence is not in NRP. After fixing so that double-blast includes the query seq in the second round, 1ois is found for t66-chim7, but not t66-chim20 or t66-chim. t66-chim7 1ois -6.377 The 1ois PDB file has three unresolved residues right at the break between the two parts of the chimeric sequence. Using the "mixed" sequence and target98 alignments gives better alignment, with the t66-chim7-1ois-global alignment looking best (though some tweaking still possible---see 1ois/t66-chim7-global-hand3.a2m) Mon Jul 27 16:31:00 PDT 1998 Kevin Karplus I am running a search now with t66_7.seq (t64+7X+t65). Blast top hit (2ori[LR]) -9.5 (basically all t64 hits, until 1ois -2.88) Double-blast pulls in 1auz and 1buz (-4.42), but cuts off before 1ois. t66_7.t98_6 gets just the t64 hits. The template library gets 1adr -49.690 1r69 -44.320 2or1L -44.110 1lliA -12.000 1ois -7.730 2liv -6.570 I hand-edited the most promising of the 1adr and 1ois alignments, and got ones that seem quite consistent to me: t66_7-1adr-hand.a2m 1ois-t66_7-hand.a2m These can, in fact, be spliced together at the join between the domains by continuing the helix, and this seems to make a consistent 3D structure. I'm not sure what to do about secondary structure prediction yet. From cline@cse.ucsc.edu Thu Jul 30 18:15:08 1998 Return-Path: cline@cse.ucsc.edu From: "Melissa Cline" Date: Thu, 30 Jul 1998 18:15:07 -0700 In-Reply-To: Kevin Karplus "Re: Questions on T64, 65, and 66" (Jul 29, 5:22pm) References: <199807300022.RAA26868@purr.cse.ucsc.edu> X-Mailer: Z-Mail (3.2.1 6apr95 MediaMail) To: Kevin Karplus Subject: Re: Questions on T64, 65, and 66 Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii > T66 IS t64+t65. I don't know what format they want it in. Perhaps > that is worth asking them. I suspect that we should just submit the > methods and results there, with no model (pointing to the models in > T0064 and T0065). Here's what the target page for Target 66 says: This is a complex of SinR (chain A) and SinI (chain B). Predictions in all formats for individual chains should be submitted as either T0064 or T0065. Only predictions of interchain residue-residue contacts and atomic coordinate predictions (in TS format) for the complex (consisting of both chains) should be submitted as T0066. So, maybe T66 is a docking target. T64 and T65 are submitted and acknowledged. From karplus@cse.ucsc.edu Thu Jul 30 20:40:19 1998 Return-Path: karplus@cse.ucsc.edu Date: Thu, 30 Jul 1998 20:40:18 -0700 From: Kevin Karplus To: venc@september.llnl.gov Cc: karplus@cse.ucsc.edu In-reply-to: <199807310153.SAA07562@september.llnl.gov> (venc@september.llnl.gov) Subject: Re: Update for target T0068 Thank you for the clarification. The 1RMG homology is not weak---it is extremely strong. Only a poor set of default parameters causes NCBI blast to miss it. All the other tools we tried (including default wu-blast) found the homology with no difficulty. There is some difficulty in getting a good alignment, so t0068 is still an excellent fold-recognition target. I have question on t0066. I have a prediction of the interaction of t0064 and t0065 which consists of aligning them both to the same structure, but I see no way to submit this as a prediction with the current format. Since the prediction is the same as for the chains separately, should I just submit a file with METHOD but no MODEL for t0066? Please advise me on how to submit this. (We've already submitted T0064 and T0065, if you need to look at our submissions to figure out what I'm talking about.) Kevin Karplus From venc@september.llnl.gov Fri Jul 31 11:46:14 1998 Return-Path: venc@september.llnl.gov From: "Ceslovas Venclovas" Date: Fri, 31 Jul 1998 11:45:27 -0700 In-Reply-To: Kevin Karplus "Re: Update for target T0068" (Jul 30, 8:40pm) References: <199807310340.UAA01683@purr.cse.ucsc.edu> X-Mailer: Z-Mail (3.2.3 08feb96 MediaMail) To: Kevin Karplus Subject: Re: Update for target T0068 Cc: venc@september.llnl.gov, adamz@sb9.llnl.gov Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii On Jul 30, 8:40pm, Kevin Karplus wrote: > Subject: Re: Update for target T0068 > Thank you for the clarification. > > The 1RMG homology is not weak---it is extremely strong. Only a poor > set of default parameters causes NCBI blast to miss it. All the other > tools we tried (including default wu-blast) found the homology with no > difficulty. > > There is some difficulty in getting a good alignment, so t0068 is > still an excellent fold-recognition target. > > I have question on t0066. I have a prediction of the interaction of > t0064 and t0065 which consists of aligning them both to the same > structure, but I see no way to submit this as a prediction with the > current format. Since the prediction is the same as for the chains > separately, should I just submit a file with METHOD but no MODEL for t0066? > > Please advise me on how to submit this. (We've already submitted > T0064 and T0065, if you need to look at our submissions to figure out > what I'm talking about.) > > Kevin Karplus >-- End of excerpt from Kevin Karplus Dear Dr. Karplus, You are right. Currently only predictions containing 3D coordinates or residue-residue contacts could be submitted for T0066 (complex of two chains). We understand the problem and will try to solve it as soon as possible. I will send you a notification message when we will be prepared to accept predictions in sequence alignment format for the complex of two chains. Best regards, Ceslovas Venclovas -- -- 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 Thu Aug 6 14:40:05 1998 Return-Path: venc@september.llnl.gov Date: Thu, 6 Aug 1998 14:39:14 -0700 From: venc@september.llnl.gov (Ceslovas Venclovas) To: karplus@cse.ucsc.edu Subject: Update for targets T0064-T0066 Dear Predictor, This is an update on two issues: 1. Two new targets (T0078 and T0079) were included in the CASP3 target database. 2. Some structural information regarding structure SinR and SinI (targets T0064, T0065, T0066) has been released in the form of an abstract. According to the authors this was the only structural information publicized. The abstract is appended below and is also available from the target descriptions on CASP3 web page at: http://PredictionCenter.llnl.gov/casp3/Casp3.html ---------------------------------------------------------------- THE CRYSTAL STRUCTURE OF SINR, AN INHIBITOR OF SPORULATION IN BACILLUS SUBTILIS IN COMPLEX WITH ITS ANTAGONIST SINI R.J. Lewis, J.A. Brannigan, W. Offen, I. Smith & A.J. Wilkinson Department of Chemistry, University of York, York. U.K. _________________________________________________________________ Amongst a panoply of molecular components which regulate the entry of Bacillus subtilis into the sporulation pathway are the proteins of the sin (sporulation inhibition) operon. The sin region encodes two proteins SinR (Mr = 14,000) and SinI (Mr = 6,000). SinR is a DNA binding protein which negatively regulates the transcription of genes, including spo0A, spoIIA, spoIIE and spoIIG, in late phase Bacillus growth. The effects exerted by SinR must be overcome in order for the sporulation process to proceed. SinI, which is encoded on the same operon as SinR but subject to different regulatory signals, acts as an antagonist of SinR through direct binding of the two proteins. This interaction involves an interesting subunit reorganisation; equilibrium sedimentation experiments indicate that SinR is a tetramer and that SinI as also multimeric. In contrast, the SinR-SinI complex is a heterodimer. We have recently solved the crystal structure of SinR in complex with SinI to 1.9 Å spacing. The phases were calculated using data collected from two selenium derivatives, in the first of these SinR contained seleno- methionine, in the second of these SinI contained selenomethionine. In the crystal, the SinI-SinR complex is, as expected, a heterodimer. SinR consists of an N-terminal 66 residue domain with very high structural similarity to the DNA binding domain of bacteriophage 434 and [lambda] CI (repressor) proteins including a helix-turn-helix DNA binding motif. A somewhat disordered region of 10 or so residues connects this domain to the C-terminal 40 residues which are involved in interactions with SinI. The interactions between SinI and SinR appear to be strong with a pair of helices from each partner interdigitating to form a significant hydrophobic core. Most interestingly, it has been noted that the central 30 residues of SinI and the 30 residues towards the C-terminus of SinR have closer sequence similarity to one another than either has to any other sequence in the database. This suggests that the interactions among SinR monomers in the tetramer may be akin to those between SinR and SinI in the heterodimer. ---------------------------------------------------------------- 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 venclovas1@llnl.gov Tue Aug 11 10:40:13 1998 Return-Path: venclovas1@llnl.gov Sender: venc@llnl.gov Date: Tue, 11 Aug 1998 10:39:19 -0700 From: Ceslovas Venclovas Organization: Lawrence Livermore National Laboratory X-Mailer: Mozilla 4.04 [en] (X11; I; IRIX64 6.2 IP28) MIME-Version: 1.0 To: karplus@cse.ucsc.edu CC: Ceslovas Venclovas , adamz@sb9.llnl.gov Subject: Submitting predictions for T0066 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear Dr. Karplus, We have realized that submitting alignment prediction for multichain target requires significant changes in the format. That could make many people unhappy. Therefore we decided not to change anything in the format but instead to provide means for converting alignment (AL) format into 3D (TS) format. The server for converting AL format to TS format is on our web site in "Local services": http://PredictionCenter.llnl.gov/local/al2ts/ Now alignments for individual chains can be converted to tertiary structure separately and combined into single model afterwards. We believe that this server provides a general solution of problems arising from imperfection of alignment format. In addition it provides a possibility for predictors to inspect their models at the 3D level before depositing. If you have some questions regarding this server please don't hesitate to contact us. Best wishes, Ceslovas Venclovas -- Protein Structure Prediction Center, Lawrence Livermore National Laboratory, Livermore, CA 94550 E-mail: venclovas1@llnl.gov Phone: (925) 422-3097 Fax: (925) 423-3608 Wed Aug 12 14:00:18 PDT 1998 I ran the tool on the web page and got 3d coordinates. There is no trick to superimposing the two 1ois predictions, but I don't know any way to get the 1adr prediction into the same coordinate system. From compbio-request Wed Aug 12 14:00:12 1998 Return-Path: karplus@cse.ucsc.edu Date: Wed, 12 Aug 1998 14:00:00 -0700 From: Kevin Karplus To: compbio@cse.ucsc.edu Subject: 3D editing tool needed! I have 3D predictions for t64 and t65, based on our alignments, using the tool at http://PredictionCenter.llnl.gov/local/al2ts/ I can superimpose those parts that come from the same PDB file, but I don't know how to paste in the other prediction, since it is in a different coordinate system. What I need is a tool that gives me a graphical view like rasmol, but allows me to move and rotate part of the structure (the domain from the other PDB file) independently of the rest, then output a PDB file. Do we have such a tool anywhere? Is there one readily available on the net? Kevin Fri Aug 14 14:07:29 PDT 1998 Kevin Karplus I spent 2 hours this morning using Insight II (courtesy of Todd Wipke) to try to patch the 1adr and 1ois alignments together. I had to fudge a bit to avoid conflicts, and am not sure I have it quite right, but 3d/try2.pdb is probably the best I'll do.