Fri Jun 30 09:42:14 PDT 2006 T0356 Make started Fri Jun 30 09:45:21 PDT 2006 Running on shaw.cse.ucsc.edu Fri Jun 30 10:02:31 PDT 2006 Kevin Karplus Modest hit by BLAST in PDB: 1ykeB 38% over 58 residues, E-value 0.004 Sat Jul 1 04:03:09 PDT 2006 Kevin Karplus No strong hits with HMMs either, best is 1ejeA E-value 3.4 and top hits are all from different folds. Oops---there is a serious problem with this target---they did not capitalize the target name in the target, so there is a discrepancy between the TARGET:= line in the Makefile and the target name. This means that no alignments ended up getting used! Two choices: rerun everything with the target name capitalized in the normal way, or try to patch up all existing files. Probably easier to rerun. Make started Sat Jul 1 04:12:15 PDT 2006 Running on lopez.cse.ucsc.edu Make started Sat Jul 1 11:04:38 PDT 2006 Running on shaw.cse.ucsc.edu I forgot to remove the first try1 decoys, so they did not get remade. I have renamed decoys/ to first-try-decoys, and created a new decoys for this run. I then moved the first-try-decoys/T0356.*.pdb.gz to decoys/T0356.first-*.pdb.gz Mon Jul 10 10:09:44 PDT 2006 Grant Thiltgen Well the remade try1 looks much better. There's still not a lot of close hits on anything, so I'm going to run some subdomains. I'm not really sure where to cut it so I'm going to select 1-180, 160-340, and 320-end. Make for 1-180 started on camano. Make for 160-340 started on whidbey. Make for 320-505 started on heehaw. Tue Jul 11 14:24:12 PDT 2006 Grant Thiltgen Some of the make for 320-505 failed. I'm not sure why. I restarted the make piping it to the make2.log file. Make started Thu Jul 13 14:25:50 PDT 2006 Running on vashon.cse.ucsc.edu Thu Jul 13 14:26:49 PDT 2006 I re-ran the make on the main directory to get the sep predictions. It turns out that the subdomains don't really have much more signal than the whole protein for anything. This isn't going to make the work on this protein any easier. The closest PDB hit is 1ykeB. I'm going to check out this file and see what it looks like. 1ykeB seems to be an all helical protein. I'm pretty sure this isn't what we're looking for. I did a blast search: RID: 1152826370-28476-155261387043.BLASTQ4 It says that it matches to pfam01977, UbiD, 3-octaprenyl-4-hydroxybenzoate carboxy-lysase. This family has been characterised as 3-octaprenyl-4- hydroxybenzoate carboxy-lyase enzymes. This enzyme catalyses the third reaction in ubiquinone biosynthesis. For optimal activity the carboxy-lase was shown to require Mn2+.. This is probably metal binding somewhere then as well. I'm looking to see if there are structures of this kind of protein anywhere. I found an article about an FMN-dependant UbiX-like decarboxylase (Pad1) in E.coli that is a paralog of UbiX, (3-octaprenyl-4-hydroxybenzoate carboxylyase, 51% sequence identity), so I am going to look at see how well it matches to the target. Unfortunately, it only matches the first 300 residues of the protein. I may add it to top hits and see how the alignment looks from SAM, just for something to see. Thu Jul 13 15:37:10 PDT 2006 Grant Thiltgen The SAM alignment got nothing. I'm not sure I'm on the right track here... I think I am going to try with the first domain and see if I can piece together something that looks vaguely protein-like from the secondary structure predictions. Thu Jul 13 17:05:27 PDT 2006 Firas Khatib I am running try2 for T160-F340 on lopez this is the difference between try1 and try2: < include rr.constraints --- > #include rr.constraints > > Constraint T164.CB V219.CB -10. 7.0 14.0 2 > Constraint V219.CB F304.CB -10. 7.0 14.0 2 > Constraint T164.CB L176.CB -10. 7.0 14.0 2 > Constraint M191.CB V219.CB -10. 7.0 14.0 0.45184803122 bonus > Constraint V219.CB I228.CB -10. 7.0 14.0 0.355821710887 bonus > Constraint V219.CB A231.CB -10. 7.0 14.0 0.327875402128 bonus > Constraint V219.CB L273.CB -10. 7.0 14.0 0.325646547231 bonus > Constraint V219.CB L229.CB -10. 7.0 14.0 0.312902262544 bonus > > Constraint T166.CB G177.CA -10. 7.0 14.0 1 > Constraint M191.CB V217.CB -10. 7.0 14.0 1 > Constraint Y179.CB M191.CB -10. 7.0 14.0 1 Fri Jul 14 17:58:22 PDT 2006 Grant Thiltgen I'm started try2 for the M1-R180 region of the protein. I made sheetconstraints, helixconstraints, and hbonds based on the secondary structure predictions. I am hoping to get something decent out of this run, but we'll see. Try2 started on whidbey. # New constraints from Secondary Structure Predictions: HelixConstraint (T0356)D8 (T0356)Q18 1 HelixConstraint (T0356)L33 (T0356)R44 1 HelixConstraint (T0356)P70 (T0356)G76 1 HelixConstraint (T0356)V82 (T0356)L95 1 HelixConstraint (T0356)K102 (T0356)K110 1 HelixConstraint (T0356)P112 (T0356)L118 1 # From sep alphabets: Hbond (T0356)L22.N (T0356)L16.O 3 Hbond (T0356)E21.N (T0356)L16.O 5 Hbond (T0356)G47.N (T0356)T42.O 5 Hbond (T0356)Q79.N (T0356)A74.O 5 Hbond (T0356)N175.N (T0356)T166.O 5 Hbond (T0356)G168.N (T0356)R173.O 5 Hbond (T0356)T166.N (T0356)N175.O 5 Hbond (T0356)M75.O (T0356)G78.N 5 Hbond (T0356)D141.O (T0356)R144.N 5 Hbond (T0356)W151.O (T0356)D154.N 5 SheetConstraint (T0356)L163 (T0356)R167 (T0356)I178 (T0356)Q174 hbond (T0356)N175 10 SheetConstraint (T0356)L50 (T0356)N54 (T0356)C64 (T0356)M60 hbond (T0356)L51 10 SheetConstraint (T0356)K123 (T0356)L125 (T0356)I134 (T0356)Q132 hbond (T0356)K123 10 SheetConstraint (T0356)K133 (T0356)V135 (T0356)C150 (T0356)M148 hbond (T0356)K133 10 SheetConstraint (T0356)I147 (T0356)C150 (T0356)T166 (T0356)L163 hbond (T0356)I147 10 StrandConstraint (T0356)L22 (T0356)T26 10 Mon Jul 17 11:52:35 PDT 2006 Grant Thiltgen Try2 for M1-R180 looks kinda crappy, but I think I can add some more constraints in to make it better. Some of the strands are flopping around and I am going to try to pull them in with the new run. I added the new constraints, and I'm starting try3 for M1-R180 on vashon. Tue Jul 18 13:07:10 PDT 2006 Grant Thiltgen I'm re-running the make on the main directory and the three subdomains. Make started Tue Jul 18 13:09:09 PDT 2006 Running on whidbey.cse.ucsc.edu Tue Jul 18 15:33:33 PDT 2006 Firas Khatib for T160-F340: try3 did not make the sheet-constraints that I wanted so I am running try4 with stronger weights on shaw. If that fails I will resort to Proteinshop Wed Jul 19 16:03:48 PDT 2006 Grant Thiltgen I need to do some work on the third subdomain. I'm not sure how well the models are going to be for the preliminary submission. We'll attempt to make a chimera of our three subdomains, submit the first undertaker run, and look at the server models to see which of them are best for the preliminary submission on Friday. I had to download the server models today. Firas then scored them with the unconstrained costfcn. Wed Jul 19 20:13:25 PDT 2006 Grant Thiltgen The best server model is the SAM_T06 server model 1. The next is Raptoress model 5. I am going to start two new undertaker runs of the whole protein and some polishing runs of some of the server models for the preliminary results on Friday. try2 and try3 are just regular undertaker runs without the rr.constraints but with the dssp-ehl2 and sheets constraints included. try2 started on shaw try3 started on lopez Wed Jul 19 21:07:55 PDT 2006 Grant Thiltgen Try4 is a polish of Sam T06 server model 1, since it was the best scoring server model. Try5 is a polish of Raptoress models 4 and 5 because they were the next two high scoring server models. try4 started on lopez try5 started on whidbey Thu Jul 20 16:24:57 PDT 2006 Firas Khatib I will recap the 5 different whole-protein runs for the preliminary submission tonight: alignments: try1 -> 1q7z, 1v3h, 1wch, and 1o2d try2 -> 1q7z, 1v3h, 1wch, and 1o2d try3 -> 1e9r, 1q6c, 1wqa, 1oz9, and 1wch try4 -> SAM_T06_server_TS1.pdb try5 -> RAPTORESS_TS4.pdb & RAPTORESS_TS5.pdb (it picked TS4 most) scoring, in all cases it's opt2 (and in most cases it's opt2-repack) try5-repack > try1-rpk > try4-rpk > try2-rpk > try3-rpk [try5 & try4 costfcn] try3-repack > try1-rpk > try2-rpk > try4 > try5-rpk [try1,try2 & try3 costfcn] try1-repack > try5-rpk > try4-rpk > try2-rpk > try3-rpk [unconstrained cstfcn] just from that, my order would be: try5-opt2-repack -- a polish of RAPTORESS_TS4 & RAPTORESS_TS5 try1-opt2-repack try4-opt2-repack -- a polish of SAM_T06_server_TS1 try3-opt2-repack -- run without rr.constraints try2-opt2-repack -- run without rr.constraints keeping in mind that once Pinal does her runs on the 3rd subdomain, we will be able to create many different looking chimeras combining the best of M1-R180, T160-F340, and T320-H505 Thu Jul 20 20:20:16 PDT 2006 Kevin Karplus Preliminary submission done with comment: Model 1 is try5-opt2.gromacs0.repack-nonPC, a reoptimization of server model RAPTORESS_YS4 by undertaker, then gromacs, then sidechain repacking by rosetta. It is rosetta's favorite of the backbones repacked so far. Model 2 is try1-opt2.repack-nonPC, a model automatically generated by undertaker, then with sidechains repacked by rosetta. It scores best with undertaker's unconstrained costfcn. Model 3 is try4-opt2.repack-nonPC, a reoptimization of our own server model (SAM_T06_server_TS1) by undertaker, then with the sidechains repacked by rosetta. Model 4 is try3-opt2.repack-nonPC, optimized from the alignments, but without any residue-residue constraints. Model 5 is try2-opt2.repack-nonPC, optimized from the alignments, but without any residue-residue constraints. We plan to work on parts of this protein independently, then put together a chimera to optimize, but hav not had time yet. ___________________________________________________________ Fri Jul 21 11:51:48 PDT 2006 Pinal Kanabar Working on third domain T320-H505 Observations : Summary.html : There no good hit for this domain. template 1ykeB has evalue 10^-2. T320-H505/best-models.pdb.gz : T0356.try1-opt2.pdb is the only model which seems to be complete. Other models have just fragments ... - resi. 500-505 histine - has his tag - there one more histidine present at 480 - there are two cycstine at resi : 401 and 361 : looking at the model they are really far apart ( Distance CYS401A.SG-CYS361A.SG: 26.481 ) so it highly unlikely that they are invloved in forming di-sulfide bond -- unless protein is folded in the different orientation. - conserved_t06 : VAL 405 ILE 351 ALA 455 ASP 402 - conserved_t2k : ARG 420 ASP 426 ALA 455 VAL 398 - conserved_t04 : GLU 328 ILE 351 CYs 401 VAL 405 ALA 455 ASP 439 we can see that very few resides are conserved. Looking with near script there are many parts which really wants to be buried. So some parts might get buried when we do chimera. As there was nothing to start with I did vast serach for my sub-domain(length 186). VAST Results : It considers top 75 resi and last 18 residues as one domain and middle 80 residues as another domain Best top structural similar template : top 75 resi : 1X7V 1VK8 1MLI middle domain : 1WQA 3PMG Mon Jul 24 11:03:03 PDT 2006 Pinal Kanabar I have created new subdomain T320-S496 after removing His tag running undertaker to see if I find anything. For subdomain T320-H505 I am going to run try2 with high wts on dry. There some parts which really needs to be buried... After this I will try working in Proteinshop and turn the sheet (365-369) which really want to be buried... Mon Jul 24 14:27:39 PDT 2006 Pinal Kanabar I started try2 with following changes : try2.costfcn: dry5 20 dry6.5 30 dry8 20 dry12 8.5 // T0356.dssp-ehl2.constraints HelixConstraint A330 L332 0.6013 HelixConstraint V334 K346 0.7181 StrandConstraint I351 Y355 0.6505 StrandConstraint L365 I370 0.7272 HelixConstraint G376 L389 0.6358 StrandConstraint F397 C401 0.8535 HelixConstraint W410 T419 0.7032 StrandConstraint T427 V430 0.6541 StrandConstraint M451 D454 0.7899 HelixConstraint P475 L488 0.626 try2 is running whidbey I edited :T0356.try1-opt2.pdb in protein shop and turned the sheet (365-369) and was hoping that I will be able to pack against helix (376-391), but this adventure was not very succefull as I was unable to pack it againt the helix which I wanted to. I thought I will still see what happens so for try3 following changes were made try3.under InfilePrefix decoys/Proteinshop/ ReadConformPDB try1_opt2_sheetheliexNear.pdb include read-pdb.unde try3.costfcn dry5 20 dry6.5 30 dry8 20 dry12 8.5 with T0356.dssp-ehl2.constraints and //Sheet Constraints #sheet constraints from conformation 0 T0356-1wqaA-t06-local-str2+near-backbone-11-0.8+0.6+0.8-adpstyle5.a2m:1wqaA SheetConstraint (T0356)T395 (T0356)V400 (T0356)R425 (T0356)V430 hbond (T0356)F397 0.512245 SheetConstraint (T0356)F397 (T0356)C401 (T0356)S449 (T0356)L453 hbond (T0356)V398 0.512245 SheetConstraint (T0356)V400 (T0356)C401 (T0356)E431 (T0356)N432 hbond (T0356)C401 0.512245 try3 is running on vashon Mon Jul 24 16:19:52 PDT 2006 Firas Khatib so I had to resort to proteinshop, which seg faulted on me TWICE while working on it today!! fromTry2-fixNtermSheets.renum.pdb manages to look somewhat like what I hope it will, so I will try inputting that to Undertaker and we'll see what happens! this will be try5 running on shaw with try4's costfcn also inputted it as try6, running on whidbey using try2's costfcn also inputted it as try7, running on whidbey using try1's costfcn Tue Jul 25 13:37:56 PDT 2006 Firas Khatib well, the strands are all in the right area, but didn't form nice sheets... time to put in more strict sheet constraints! Wed Jul 26 16:38:30 PDT 2006 Pinal Kanabar I like the new subdomain T320-S496.No new hit is found. try1 looks slight better than the earlier domain (T320-H505 with his tag) Removing his tag gave little more room to align all the sheets. I with the help of Grant took T0356.try1-opt2.pdb and we used Proteinshop to align the sheets - 356.try1-opt2_alignsheet.pdb. try2 is running on camano with following changes try2.under InfilePrefix decoys/Proteinshop/ ReadConformPDB T0356.try1-opt2_alignsheet.pdb try2.costfcn # From T0356.t06.str2.constraints HelixConstraint A330 Q345 0.752 StrandConstraint I351 Y355 0.676 StrandConstraint L365 K371 0.719 HelixConstraint H377 F388 0.736 StrandConstraint F397 C401 0.85 HelixConstraint W410 T419 0.705 StrandConstraint V428 L429 0.822 StrandConstraint M451 D454 0.746 HelixConstraint P475 L488 0.63 #Sheet constrains from : try1-opt2.sheets SheetConstraint (T0356)E350 (T0356)L356 (T0356)I370 (T0356)R364 hbond (T0356)E350 20 SheetConstraint (T0356)F397 (T0356)D402 (T0356)R425 (T0356)V430 hbond (T0356)I399 20 SheetConstraint (T0356)V398 (T0356)D403 (T0356)G448 (T0356)L453 hbond (T0356)V400 20 SheetConstraint (T0356)D402 (T0356)D403 (T0356)E431 (T0356)N432 hbond (T0356)D403 10 #new SheetConstraint SheetConstraint (T0356)V428 (T0356)R425 (T0356)V367 (T0356)R364 hbond (T0356)V367 50 #changes :Added :T0356.t06.str2.constraints, Sheet constrains from : try1-opt2.sheets # from to #constraints 10 50 #StrandConstraint F397 C401 0.916 0.85 try2 is running on camano Thu Jul 27 12:51:50 PDT 2006 Grant Thiltgen For the M1-R180 subdomain, I am attempting to move a strand from try2 next to the sheet in order to add on to the first sheet. I am also increasing the weights on the hbonds, sheets, and helices. This is try4 and it is running on shaw. Thu Jul 27 14:42:09 PDT 2006 Pinal Kanabar Domain: T320-S496 T0356.try1-opt2.gromacs0.repack-nonPC.pdb.gz is Rosetta's favourite try2 looks good it did hold new sheets together , but I dont see Hbond between them So I try to optimize T0356.try2-opt2.pdb.gz and get Hbond Thu Jul 27 14:55:48 PDT 2006 Grant Thiltgen Try4 for the subdomain looks pretty crappy. I think I need to use the constraints from the model and not what I put in for try5, and then add in the constraint for the extra strand. I took the constraints from try2 and took out the helices that I didn't want and added in the strands that I did. This is try5 and started on shaw. Thu Jul 27 15:35:34 PDT 2006 Pinal Kanabar Domain: T320-S496 optimizing using :T0356.try2-opt2.pdb.gz try3.costfcn : same as try2 with following changes. #new SheetConstraint SheetConstraint (T0356)R425 (T0356)V430 (T0356)R364 (T0356)T369 hbond (T0356)V367 60 #new changes #StrandConstraint R425 N432 0.822 try3 is running on camano Fri Jul 28 10:29:44 PDT 2006 Pinal Kanabar In try3 I managed to get few more hbond between the strands .. try3 scores best with score-all+servers.unconstrained.pretty and it also rosetta's favourit ... try1 and try3 are my best try's till now they have different arrangment of sheets. try1-> try1.log.gz -> 1wqaA ,1bqk, 1amy, 1vp4A try3-> T0356.try2-opt2.pdb.gz ->T0356.try1-opt2_alignsheet.pdb (proteinshop) Sun Jul 30 15:43:23 PDT 2006 Firas Khatib I will try to add sheet constraints to the models in the second domain. try5 scores best (of the proteinshopped models) maybe I will start from that one. of the proteinshopped models, I think that try6 has no hope to be salvaged. try7 might have a chance, but try5 is the best bet. I like the sheet that formed from I271-I277 with V303-R312 since it is consistent with dssp-ehl2 & str2. I want to make that very strong and then try to form sheets with the strands on either side. Sun Jul 30 16:21:51 PDT 2006 Firas Khatib try8 is running on shaw for domain T160-F340, using try5 as input. Sun Jul 30 21:59:15 PDT 2006 Firas Khatib well, try8 scores better than try5 and the proteinshop model they came from. I do like what it did with helix S240-R249, but now the core is terribly foamy since the strands didn't form sheets well. Mon Jul 31 11:48:21 PDT 2006 Grant Thiltgen I am going to try to get the sheet to form by using ProteinShop. I am proteinshopping try2-opt2 of the M1-A180 subdomain to get the sheet to form. I have the ProteinShopped model, try2-opt2-moved-strand. There are still some helices in it that I want as strands, but I want to at least get this formed first, and we'll see about the rest. Try6 in the M1-A180 region is going to attempt to keep that strand and form the sheet. Try6 started on whidbey. Mon Jul 31 15:06:14 PDT 2006 Grant Thiltgen Try6 scores best now. The sheet isn't formed as great as I would like but it's there. I'm going to work on making a chimera of the three best models from the subdomains. I am also going to start a run for the entire protein using constraints from all three sections. Mon Jul 31 16:06:00 PDT 2006 Grant Thiltgen Okay, so I made a chimera and attempted to proteinshop things to get some of the overlapping regions out of the way. I think I did an all right job. I also made try6.costfcn which has the constraints from all three domains. The first chimera model is try6 from the first domain, try8 from the second domain, and try3 from the third domain. [Mon Jul 31 16:28:09 PDT 2006 Kevin Karplus Exactly where is the crossover between different models when making the chimera? What is the name for the resulting chimera? Please be more explicit in keeping lab notebook notes. Particularly for chimeras, it can be difficult to reconstruct where they came from. ] I am also running the try6.costfcn with try7. Try7 is going to attempt to make it all from alignments instead of polishing a chimera. We'll see what differences I get. Both try6 and try7 started on whidbey. I'll run try8 and try9 starting from the beginning and using the same costfcn from try7 as well, to see if we get something different. I will start these using para-trickle-make on the cluster, so they won't use some CPU's and see what we'll get. Mon Jul 31 16:37:05 PDT 2006 Grant Thiltgen For the chimera, the crossover for try6-opt2 in the first domain includes up to residue 171. Try8-opt2 for the second domain includes up to residue 331. Try3-opt2 for the third domain is the rest. The chimera is also without the HIS-tag, which I will add on later. I am running two more tries for the first subdomain. They include the try2 parameters, plus some parameters for including the first sheet. I am going to increase the sheet parameters though to try to get sheets to form. This is try7 and try8 for M1-R180. Try7 has the original weights, try8 has the weights increased for the sheets and the helices. Both try7 and try8 I sent to the cluster. \ Mon Jul 31 22:06:12 PDT 2006 Grant Thiltgen Ooops! I didn't put on the HIS-tag, and now it didn't run because it couldn't read in the file since all the residues weren't there. I need to add them on into the chimera and re-run. I tacked on the HIS-tag from try1-opt2 (residues 497-505) from try1-opt2 of the T320-H505 domain. I changed try6.under to reflect the new PDB file and restarted try6 on shaw. Tue Aug 1 08:52:12 PDT 2006 Grant Thiltgen Of the models I ran from fragments, I like try8 the most. There's a huge disordered region, but a lot of it is very good, so I need to see if I can keep those constraints and fix the N-terminal end. With the unconstrained costfcn, currently the best scoring model is the chimera model I put together. It's still running though, and try6-opt1 is the best scoring. Tue Aug 1 09:01:18 PDT 2006 Pinal Kanabar I am running try4 which is based on try1 and will try to see if I can get long stands like Robetta models # I will use T0356.t06.dssp-ehl2.constraints HelixConstraint V334 K346 0.6924 StrandConstraint I351 Y355 0.6223 #new StrandConstraint S362 K371 0.9 HelixConstraint G376 L389 0.6526 #StrandConstraint F397 C401 0.8585 StrandConstraint K396 C401 0.8585 HelixConstraint W410 T419 0.7114 StrandConstraint T427 V430 0.6614 #StrandConstraint M451 D454 0.7817 StrandConstraint S449 A455 0.9.0 HelixConstraint P475 L488 0.6352 # try1.sheet constraints with some changes SheetConstraint (T0356)E350 (T0356)L356 (T0356)K371 (T0356)L365 hbond (T0356)E350 0.65 SheetConstraint (T0356)F397 (T0356)D402 (T0356)R425 (T0356)V430 hbond (T0356)I399 0.65 SheetConstraint (T0356)V398 (T0356)D403 (T0356)G448 (T0356)L453 hbond (T0356)V400 0.65 SheetConstraint (T0356)D402 (T0356)D403 (T0356)E431 (T0356)N432 hbond (T0356)D403 0.65 try4 is running on camano Tue Aug 1 10:29:23 PDT 2006 Pinal Kanabar try3 which has 2-3 breaks with range 3.1 to 1.4 so I will do polishing run on that I made following changes try5.under : InfilePrefix decoys/ ReadConformPDB T0356.try3-opt2.pdb try5.costfcn SetCost wet6.5 15 near_backbone 5 way_back 5 dry5 15 dry6.5 30 dry8 25 dry12 6 \soft_clashes 60 backbone_clashes 2 \ break 250 \ #try3-opt2.sheets and #try3-opt2.helices I removed following helix constraints fro this as in str2 alphabets it was predicted as coil region #HelixConstraint (T0356)P434 (T0356)S442 #HelixConstraint (T0356)F440 (T0356)S445 try5 is running on chacha Tue Aug 1 10:58:13 PDT 2006 Pinal Kanabar George suggested following templates might be good so my try6 is based on this templates Subdomain 310 - 500 I think the first four are ok. Yes, it might just be a 3.40.50.1820 or there might be a lot of samples of 3.40.50.1820. [Tue Aug 1 14:26:45 PDT 2006 Kevin Karplus What the h*** does 3.40.50.1820 mean? It looks like some sort of classification system, but it doesn't match the SCOP numbering scheme. ] # program: alphabetmatch # George Shackelford # # Target: T0356 # length: 505 # length range: 177 to 210 # alphabets used: # ehl2 burial # id score per residue 5S 10N 10N 1agy 432.778 0.856987 3.40.50.1820-197 1cus 431.884 0.855216 3.40.50.1820-197 1auq 431.708 0.854867 3.40.50.410-208 1cue 430.687 0.852845 3.40.50.1820-197 1qozA 429.697 0.850885 3.40.50.1820-206 1g66A 429.481 0.850458 3.40.50.1820-207 1uexC 426.543 0.844639 3.40.50.410-202 1iag 425.559 0.842691 3.40.390.10-201 1atlA 424.835 0.841256 3.40.390.10-200 1fnsA 424.724 0.841038 3.40.50.410-196 try6.under : InfilePrefix 1agy/ include read-alignments-scwrl.under InfilePrefix 1cus/ include read-alignments-scwrl.under InfilePrefix 1auq/ include read-alignments-scwrl.under InfilePrefix 1cue/ include read-alignments-scwrl.under InfilePrefix 1qozA/ include read-alignments-scwrl.under InfilePrefix 1g66A/ include read-alignments-scwrl.under InfilePrefix 1uexC/ include read-alignments-scwrl.under InfilePrefix 1iag/ include read-alignments-scwrl.under InfilePrefix 1atlA/ include read-alignments-scwrl.under InfilePrefix 1fnsA/ include read-alignments-scwrl.under and commented out : all Include statements try6.costfcn SetCost wet6.5 15 near_backbone 5 way_back 5 dry5 15 dry6.5 25 dry8 20 dry12 6 \soft_clashes 20 backbone_clashes 2 \ break 80 \ include T0356.dssp-ehl2.constraints include T0356.undertaker-align.sheets include rr.constraints try6 is running on shaw Tue Aug 1 11:20:49 PDT 2006 Grant Thiltgen I'm not sure how well George's stuff will work, because I'm not sure how well it matches since his results aren't that intuitive. [Tue Aug 1 14:29:22 PDT 2006 Kevin Karplus His alphabetmatch method has also never been tested, and is based on ideas that have never worked at CASP in the past, so you are right to be skeptical. ] Try6 for the whole protein is still running. I am going to work on another chimera for the whole protein, but I am going to have to cut back on the interations since the previous is taking over 14 hours to complete. For the 2nd chimera, I took try1-opt2 of M1-R180 and Try2-opt2 of T160-F340 and included up to residue 171 of the first model and the rest of the second. I then used try1-opt2 of the third domain, T340-S496 and included up to residue 331 of the 2nd domain and the rest of the third domain. I then took try1-opt2 of the T340-H505 domain and added the HIS-tag. I then sent it to undertaker for optimization, but I cut down some of the iterations of the program for time problems. This is try10 and try10 is started on whidbey. Tue Aug 1 14:08:48 PDT 2006 Grant Thiltgen I am building one more chimera. The third chimera is taking try1-opt2 of M1-R180, up to residue 160. The second region is taking residues 161-330 of try5-opt2 of T160-F340. The third region is taking residues 331-496 of the try5-opt2 of T320-S496. I added a HIS-tag from try1-opt2 of T320-H505. I am going to optimize this third chimera twice. The first with the constraints from the models, the second including some h-bond constraints. Try11 (without Hbonds) started on shaw. Try12 (with Hbonds) started on whidbey. Both try11 and try12 are using the 3rd chimera model. Tue Aug 1 15:50:24 PDT 2006 Grant Thiltgen The 2nd chimera blew up in undertaker. There were too many backbone clashes. When trying to fix it with proteinshop, it turns out the model is two overlapped in that area to keep any of the secondary structure elements intact. I think that that chimera isn't salvageable, but I have more models of the first one running. I am hoping they will be finished early enough, if they aren't, I can do a submission later tonight. I am going to launch a few more runs with the first chimera model too in order to get some variety. I am hoping that in trying to optimize things undertaker is going to move stuff around differently. Try13 and try14 are based on the first chimera model using the helices and sheets from the models in each of the regions. I am including hbond constraints in try13 and removing them in try14. Tue Aug 1 18:21:11 PDT 2006 Pinal Kanabar I was not sure recap for my domain was necessary or not .... but in case T320-S496 try1-> try1.log.gz -> 1wqaA ,1bqk, 1amy, 1vp4A try2-> T0356.try1-opt2_alignsheet.pdb (Protein shop model) try3-> T0356.try2-opt2.pdb.gz ->T0356.try1-opt2_alignsheet.pdb (proteinshop) try4-> T0356.try1-opt2.pdb -> 1wqaA ,1bqk, 1amy, 1vp4A try5-> T0356.try3-opt2.pdb > T0356.try2-opt2.pdb.gz ->T0356.try1-opt2_alignsheet.pdb (proteinshop) try6-> All new templates suggested by George. [Note This try is not used at all] try7-> T0356.try4-opt2.pdb -> T0356.try1-opt2.pdb -> 1wqaA ,1bqk, 1amy, 1vp4A try1 and try3 are 2 different models I could get and other try's are polishing run on them. Tue Aug 1 19:00:30 PDT 2006 Grant Thiltgen The runs are almost finished. Currently, I believe our top model is probably try6-opt2. So for the submission I have the following: T0356.try6-opt2.pdb T0356.try11-opt2.repack-nonPC.pdb T0356.try13-opt2.pdb T0356.try5-opt2.gromacs0.repack-nonPC.pdb T0356.try1-opt2.pdb T0356.try6-opt2.pdb is a model created from a chimera of three subdomains. The subdomains we created were M1-R180, T160-F340, and T320-S496. We chose to remove the HIS-tag for creating our subdomain models because it was getting in the way of some of our models. The chimera we made for this model used try6-opt2 of the first subdomain, try8-opt2 of the second subdomain and try3-opt of the third subdomain. Try6-opt2 was based on try2-opt2, which used some hbond constraints and ehl2 constraints to create sheets and helices by hand. Then I used proteinshop to add a third strand to the first sheet which was trailing off by itself. Try8-opt2 was based on a proteinshopped model of try2 which needed some sheets fixed. Try3-opt2 is based off of try1-opt2 of the third domain. It was the original undertaker run that had been proteinshopped to get all the strands to form a sheet. I used up to residue 171 of the first domain, residue 331 of the second domain, and the rest of the third domain, with a HIS-tag tacked onto the end. We used protein shop to move some of the clashing secondary structure and backbone elements apart from each other. I then sent the chimera into undertaker for a full run (which ran close to 15 hours) to get this model. T0356.try11-opt2.repack-nonPC.pdb is our second highest scoring undertaker model. It uses a different chimera this time. For this chimera, we used try1-opt2 of the first domain, try5-opt2 of the second domain, and try5-opt2 of the third domain. Try1-opt2 was the first model from the undertaker run of the first domain. Try5-opt2 is based on try2 of the second domain with some proteinshopped strands to form a sheet. Try5-opt2 is a polished model of try3-opt2 which has the proteinshopped strands. I cut the first domain at residue 160, the second domain at residue 330 and the rest from the third model. I then added a HIS-tag. We used protein shop to move some of the clashing secondary structure and backbone elements apart from each other. I sent it to undertaker with the constraints from each of the subdomains to keep the secondary structure elements, and this is the result. T0356.try13-opt2.pdb is based on the first chimera again using the constraints from the three individual subdomains and with adding hbond constraints from the predicted hbonds from the sep and notor alphabets. T0356.try5-opt2.gromacs0.repack-nonPC.pdb is the best scoring model using the rosetta scoring function. It was a polish of RAPTORESS_TS4 and RAPTORESS_TS5 which were high scoring server models. It mostly used TS4 in building this model. T0356.try1-opt2.pdb is the original undertaker run starting from scratch. ----- Unfortunately this is all we have. I built another chimera, but I couldn't get the second and third subdomains together in a way to prevent clashing. I tried running it through undertaker anyway and the results were try10-opt2, which basically pulled the entire protein apart. Tue Aug 1 20:14:52 PDT 2006 Kevin Karplus Submitted with comment We did not have any good fold-recognition hits for T0356---neither as a full chain nor broken up into putative subdomains. The protein is a bit too large for ab initio techniques, but we threshed around with it for a while anyway. We got some sheet fragments that look pretty good, but no overall fold. Model 1, T0356.try6-opt2.pdb, is a model created from a chimera of three subdomains. The subdomains we created were M1-R180, T160-F340, and T320-S496. We chose to remove the HIS-tag for creating our subdomain models because it was getting in the way of some of our models. The chimera we made for this model used try6-opt2 of the first subdomain, try8-opt2 of the second subdomain and try3-opt of the third subdomain. Try6-opt2 was based on try2-opt2, which used some hbond constraints and ehl2 constraints to create sheets and helices by hand. Then Grant used proteinshop to add a third strand to the first sheet which was trailing off by itself. Try8-opt2 was based on a proteinshopped model of try2 which needed some sheets fixed. Try3-opt2 is based off of try1-opt2 of the third domain. It was the original undertaker run that had been proteinshopped to get all the strands to form a sheet. Grant used up to residue 171 of the first domain, residue 331 of the second domain, and the rest of the third domain, with a HIS-tag tacked onto the end. We used ProteinShop to move some of the clashing secondary structure and backbone elements apart from each other. Grant then sent the chimera into undertaker for a full run (which ran close to 15 hours) to get this model. Model 2, T0356.try11-opt2.repack-nonPC.pdb, is our second highest scoring undertaker model. It uses a different chimera this time. For this chimera, we used try1-opt2 of the first domain, try5-opt2 of the second domain, and try5-opt2 of the third domain. Try1-opt2 was the first model from the undertaker run of the first domain. Try5-opt2 is based on try2 of the second domain with some proteinshopped strands to form a sheet. Try5-opt2 is a polished model of try3-opt2 which has the proteinshopped strands. Grant cut the first domain at residue 160, the second domain at residue 330 and the rest from the third model. I then added a HIS-tag. We used protein shop to move some of the clashing secondary structure and backbone elements apart from each other. Grant sent it to undertaker with the constraints from each of the subdomains to keep the secondary structure elements, and this is the result. Model 3, T0356.try13-opt2.pdb, is based on the first chimera again using the constraints from the three individual subdomains with hbond constraints added for the predicted hbonds from the sep and notor alphabets. Model 4 is T0356.try5-opt2.gromacs0.repack-nonPC.pdb, the best scoring model using the rosetta scoring function. It is an optimization of a reoptimization of server model RAPTORESS_TS4 by undertaker, then gromacs, then sidechain repacking by rosetta. RAPTORESS_TS4 and RAPTORESS_TS5 were the high-scoring server models with the unconstrained costfcn. Model 5 is T0356.try1-opt2.pdb, the atuomatic undertaker run starting from alignments. ---------------------------------------------------------------------- Please do a "make clean" when you are finished with this target, after you do a final submission. (Even if you decide to do no more submissions, do a "make clean" when you are finished.)