Mon Jun 5 11:51:25 PDT 2006 T0316 Make started Mon Jun 5 11:52:35 PDT 2006 Running on lopez.cse.ucsc.edu Mon Jun 5 12:45:12 PDT 2006 Kevin Karplus No strong BLAST hits (best is 1tcmA E-value 1.4) There are enough homologs in NR to make a good conservation signal (over 3500 sequences in the t06 alignment). The t06 alignment even includes some PDB files (1kh3A, 1vl2A, 1kp3A, 1gpmA, 1ni5A, 1wy5A, 1xnhA, 2d13A). Mon Jun 5 13:41:08 PDT 2006 Kevin Karplus Strong HMM hits for c.26.2.1: 1vl2A, 1wy5A, 1k92A, 1gpmA, 1xngA, 1kqpA, ... Mon Jun 5 21:04:10 PDT 2006 Kevin Karplus The alignments in undertaker-align are in agreement out to about A130, but the C-terminal end has all sorts of scatter. It might be best to do a subdomain A130-I373. Tue Jun 6 08:23:13 PDT 2006 Kevin Karplus The prediction from SAM_T06_server was reported as having lots of clashes and breaks, as does try1-opt2. The constraints are not well met in try1-opt2, possibly because they are rather inconsistent. The part that does seem consistent has three sheet constraints: SheetConstraint R8 V11 D122 A125 hbond V10 1 # maybe off by 1 SheetConstraint T7 M13 Y31 F37 hbond V9 1 SheetConstraint I34 K39 P69 N74 hbond G35 1 We may be able to extend the first domain out as far as A183. Looking at the A130-I373 predictions, there are no really strong templates. The best is probably 1xngA (Evalue 0.1), but it only picks up a helix between K176 and E185. There may be an anti-parallel barrel at the C-terminus. We should do a subdomain starting with A290 (A290-I373 started on camano). I also started M1-A183 and L171-P300 on lopez. Tue Jun 6 15:07:06 PDT 2006 Kevin Karplus When L171-P300 got to the try1 undertaker run, I killed the job on lopez and restarted it on the farm cluster, since Firas's machine shaw is down, and he is working on lopez. There are no good hits for L171-P300 (either with blast or the HMMs). This looks like it will really be an ab initio run. A290-I373 also has no good hits. M1-A183 has excellent hits (1wy5A, 1vl2A, ...) with the HMMs, but not with blast. Tue Jun 20 15:28:19 PDT 2006 Grant Thiltgen I looked in the PDB for other tRNA methyltransferases, and there was one with a structure in Streptococcus pneumoniae, but with less residues. I'm wondering if there might be a big insert in the middle of the protein, that may be throwing off the alignments. I did a pairwise alignment with CLUSTALW on the web, and there are long stretches of aligned sequences with some gaps between them, so there may indeed be similarities with just some inserts. Although there are also a lot of substitutions. I'm going to try a run with the big ~50 amino acid insert removed to see if it perhaps might find this PDB file from the alignments afterwards. Plus I'll run a SAM-T06 alignment for the target and the tRNA and see what it says. T0316_AAK74304__Streptococcus_ MSDNSKTRVVVGMSGGVDSSVTALLLKEQGYDVIGIFMKNWDDTDENGVC 1YZH_A|PDBID|CHAIN|SEQUENCE ----SNAMRVRNRKG-----ATELLEANPQYVVLNP-------------L *:: * . .* .* ** : * *:. T0316_AAK74304__Streptococcus_ TATEDYKDVVAVADQIGIPYYSVNFEKEYWDRVFEYFLAEYRAGRTPNPD 1YZH_A|PDBID|CHAIN|SEQUENCE EAKAKWRDLFGNDNPIHVEVGSG-------------KGAFVSGMAKQNPD *. .::*:.. : * : * * . . *** T0316_AAK74304__Streptococcus_ VMCNKEIKFKAFLDYAITLGADYVATGHYARVARDEDGTVHMLRGVDNGK 1YZH_A|PDBID|CHAIN|SEQUENCE INYIGIDIQKSVLSYALDKVLEVGVP-------------NIKLLWVD-GS : *:.*.**: : .. * ** *. T0316_AAK74304__Streptococcus_ DQTYFLSQLSQEQLQKTMFPLGHLEKPEVRRLAEEAGLSTAKKKDSTGIC 1YZH_A|PDBID|CHAIN|SEQUENCE DLTDYFEDGEIDRLYLNFSDPWPKKRHEKRRLTYKTFLDTFKR------- * * ::.: . ::* .: :: * ***: :: *.* *: T0316_AAK74304__Streptococcus_ FIGEKNFKNFLSNYLPAQPGRMMTVDGRDMGEHAGLMYYTIGQRGGLGIG 1YZH_A|PDBID|CHAIN|SEQUENCE -------------ILPENGEIHFKTDNRGLFEYS---LVSFSQYGMKLNG ** : :..*.*.: *:: ::.* * * T0316_AAK74304__Streptococcus_ GQHGGDNAPWFVVGKDLSKNILYVGQGFYHDSLMSTSLEASQVHFTREMP 1YZH_A|PDBID|CHAIN|SEQUENCE -------------------------------------------------- T0316_AAK74304__Streptococcus_ EEFTLECTAKFRYRQPDSKVTVHVKGEKTEVIFAEPQRAITPGQAVVFYD 1YZH_A|PDBID|CHAIN|SEQUENCE -----------------VWLDLHASDFEGNVMTEYEQKFSNKGQVIYRVE : :*... : :*: *: . **.: : T0316_AAK74304__Streptococcus_ GEECLGGGLIDNAYRDGQVCQYI 1YZH_A|PDBID|CHAIN|SEQUENCE AEF-------------------- .* I tried SAM-t06, but it's hard for me to interpret the a2m file I got as a result. I think that there's a huge insert in the same area as this alignment, but I'm not too sure. I may be barking up the wrong tree here. Tue Jun 21 08:48:38 PDT 2006 Grant Thiltgen Oops, I think that was a bad plan. Now I know. I'll probably have to try to piece something together by hand with the ab initio end. Wed Jun 21 12:55:27 PDT 2006 Grant Thiltgen I started try2 with new sheet constraints in the C-terminal region of the protein. I also included some separation of 3 hits from the o_notor2 and n_notor2 alphabets. Wed Jun 21 16:34:14 PDT 2006 Grant Thiltgen try2 looks horrible. I'm really not sure what to do with this protein. The region A130-I373 only seems to match up kind of well with 1xngA, so I'm going to run this domain with this protein for fragments. I am also running try3 on the whole target with just the alignments from 1yzhA to see if it can do something that looks like the other tRNA protein. Thu Jun 22 09:34:40 PDT 2006 Grant Thiltgen So try3 on the whole protein looks pretty bad. try2 on A130-I373 has some good things going, I need to look more into it and try to see if I can do something about the really bad region. (Around residues 190-225). I'll try to that together somehow, and make a chimera with the first half of the protein. Thu Jun 22 14:01:06 PDT 2006 Grant Thiltgen I started try2 for the first half of the protein M1-A183. I took out the automatic constraints and left in a few I liked. I saw that there was a high probability in the notor alphabets for a separation of five at the end of the helices, so I tried including Hbond constraints for those. I am also going to spend time to score the servers to see if they got anything decent. Oh, they're already scored! Nice. Robetta model 4 looks pretty decent. I am also starting try3 for A130-I373 with some more sheet, helix and Hbond constraints to see if I can clean up the bad area. I think I may even do a try4 for the whole protein with the constraints from both parts. Mostly because there is a small strand in the second region I want to pair with a strand in the first region, and I'm not sure I'll be able to do it as a chimera. Thu Jun 22 16:18:07 PDT 2006 Grant Thiltgen try2 on M1-A183 looks pretty good. The C-terminal region which was kind of strange looking in try1 got cleaned up in try2. This may be a decent one to make a chimera of if I find a good A130-I373 region. try3 on the A130-I373 region looked pretty bad. I think it's because I only ran with the fragments from 1xngA. I'm re-running with the same constraints using all the fragments. I'm starting try4 on orcas in order to try to get the sheets I want. Thu Jun 22 16:47:47 PDT 2006 Grant Thiltgen try4 on the whole protein is keeping the region at the beginning of the protein that we like. The rest is heading in the right direction. I'm going to run try5 with the same sheets, but increasing the weights to try to really get them to form. I also increased the constraints weight in the costfcn to help. Thu Jun 22 19:49:02 PDT 2006 Grant Thiltgen try4-opt2 on A130-I373 looks really good. I need to see if I can make a chimera of the two ends now. try5 for the whole protein isn't looking so good. Thu Jun 22 21:42:59 PDT 2006 Grant Thiltgen In order to get that sheet I am trying to get, I think I am going to have to do a run from residues 1-205 to try to get the sheet to form. I'll start that run tonight. Fri Jun 23 08:41:00 PDT 2006 Grant Thiltgen try 1 is finished on A1-K205. I am going to run try2 with the constraints I was looking for for the sheet I was trying to form to see if it does any better with only half the protein instead of the entire protein. I hope to get the extra little part of the parallel sheet to work. Fri Jun 23 09:13:08 PDT 2006 Grant Thiltgen I'm starting try6 for the whole protein. I built a chimera of try2 from M1-A183 and try4 of A130-I373, cutting the protein at residue 132, using residue 132 from try4 of A130-I373. try6 is going to attempt to use undertaker to clean it up a bit. We'll see how it works. I also noticed that try5 for the whole protein was only using the fragments from my failed attempt at trying to match the other tRNA transferase from the PDB, so I will start try7, with the same constraints as try5, but with the fragments for all the close matches. Fri Jun 23 11:21:53 PDT 2006 Grant Thiltgen I am starting try8 for the whole protein. I took the chimera for try6 and moved it around a bit in protein shop to alleviate some of the clashes. I hope it will work. Sat Jun 24 11:23:15 PDT 2006 Grant Thiltgen Both try6 and try8 from the chimeras look pretty good. try8 is slightly better, probably because I moved a clashing loop region away from an alpha helix. In the decoys/score-all+servers.unconstrained file, try8 scores better than try6. Then try2 and try3. The robetta models appear to do well, so I am going to run try9 as a polishing run of the robetta models. Try10 will be a polishing run of try8 and try6. I am also going to run a few more runs of the A130-I373 domain to see if I get any new interesting conformations to make into a chimera with the first domain. Sat Jun 24 11:52:00 PDT 2006 Kevin Karplus try8 looks ok, except for the region G127-L174. Do we have any models that do well in that region? Maybe we need some antiparallel sheet constraints for strands E78-D122, F127-E136, E136-D147? (Obviously, those are not the right boundaries) Sun Jun 25 10:14:05 PDT 2006 Kevin Karplus Grant did some polishing runs yesterday, but hasn't commented on them yet. With the try10=try9 costfcn, try10-opt2 scores best, and try9-opt2.repack-nonPC does fairly well, followed by try6-opt2, try8-opt2, try4-opt2.repack-nonPC. With the unconstrained costfcn, the order is try8, try10, try6 Rosetta doesn't like repacking any of the models, but try10-opt2 is the best of a bad lot. Sun Jun 25 10:29:53 PDT 2006 Kevin Karplus I think that the domain boundaries are rather vague, so I'm starting some more possible subdomains: M1-D137 (make started on whidbey) and D122-K269 (make started on shaw). Sun Jun 25 10:35:48 PDT 2006 Kevin Karplus Rather than using the subdomains to make chimeras (which may be necessary, but for which there is not much time), we may just want to pick up Sheetconstraints from the try1-opt2 (or higher) models of the subdomains. Currently we have # A130-I373/decoys/try1-opt2.sheets: SheetConstraint R131 E136 R144 T139 hbond V132 1 SheetConstraint C200 F201 T224 V225 hbond C200 1 SheetConstraint G248 H253 F261 D256 hbond I249 1 SheetConstraint I249 H253 S291 S287 hbond G250 1 SheetConstraint D266 L267 I271 N270 hbond L267 1 SheetConstraint I271 V274 L267 G264 hbond V274 1 SheetConstraint P300 T304 D350 V346 hbond E301 1 SheetConstraint E298 T304 Y313 C307 hbond E302 1 SheetConstraint K310 R314 E335 V331 hbond R312 1 SheetConstraint Q344 F348 I360 G356 hbond V346 1 # A130-I373/decoys/try2-opt2.sheets: SheetConstraint D229 H233 I241 M237 hbond G231 1 SheetConstraint G248 G251 F261 A258 hbond I249 1 SheetConstraint I249 H253 S291 S287 hbond G250 1 SheetConstraint V263 L267 V274 N270 hbond G264 1 SheetConstraint F303 L305 R338 P336 hbond T304 1 SheetConstraint L305 T308 F333 E330 hbond C307 1 SheetConstraint C307 T308 A363 N362 hbond T308 1 SheetConstraint T308 R314 D361 L355 hbond K310 1 SheetConstraint V320 K325 F333 K328 hbond T321 1 SheetConstraint Q344 F348 I360 G356 hbond V346 1 # A130-I373/decoys/try3-opt2.sheets: SheetConstraint A130 D135 R144 T139 hbond R131 1 SheetConstraint A217 Q218 M222 R221 hbond Q218 1 # A130-I373/decoys/try4-opt2.sheets: SheetConstraint A130 D135 R144 T139 hbond R131 1 SheetConstraint G220 M223 K265 V262 hbond M222 1 SheetConstraint V225 M230 Y239 A234 hbond D226 1 SheetConstraint G235 T240 I249 R244 hbond M237 1 SheetConstraint Y273 G275 I340 R338 hbond V274 1 SheetConstraint G275 L283 E335 E327 hbond G277 1 SheetConstraint Q276 Y279 Y364 D361 hbond F278 1 SheetConstraint F278 M284 D361 L355 hbond H280 1 SheetConstraint H294 T296 V324 V322 hbond F295 1 SheetConstraint E301 L305 D350 V346 hbond E302 1 SheetConstraint M299 L305 R314 T308 hbond F303 1 SheetConstraint R312 P316 E335 V331 hbond Y313 1 SheetConstraint Q344 F348 I360 G356 hbond V346 1 SheetConstraint F348 D350 L355 E353 hbond D350 1 # A130-I373/decoys/try5-opt2.sheets: SheetConstraint A130 D135 R144 T139 hbond R131 1 SheetConstraint M222 T224 V263 F261 hbond M222 1 SheetConstraint V225 M230 Y239 A234 hbond D226 1 SheetConstraint G235 T240 I249 R244 hbond M237 1 SheetConstraint L272 G275 T341 R338 hbond V274 1 SheetConstraint G275 L283 E335 E327 hbond G277 1 SheetConstraint Q276 Y279 Y364 D361 hbond F278 1 SheetConstraint F278 M284 D361 L355 hbond H280 1 SheetConstraint H294 T296 V324 V322 hbond F295 1 SheetConstraint M299 L305 R314 T308 hbond E301 1 SheetConstraint E301 L305 D350 V346 hbond E302 1 SheetConstraint R312 P316 E335 V331 hbond Y313 1 SheetConstraint Q344 F348 I360 G356 hbond V346 1 SheetConstraint F348 D350 L355 E353 hbond D350 1 # A130-I373/decoys/try6-opt2.sheets: SheetConstraint L143 R144 L171 G172 hbond L143 1 SheetConstraint R221 T224 G264 F261 hbond M222 1 SheetConstraint V225 M230 Y239 A234 hbond D226 1 SheetConstraint G235 T240 I249 R244 hbond M237 1 SheetConstraint L272 G275 T341 R338 hbond V274 1 SheetConstraint G275 L283 E335 E327 hbond G277 1 SheetConstraint Q276 Y279 Y364 D361 hbond F278 1 SheetConstraint F278 M284 D361 L355 hbond H280 1 SheetConstraint H294 R297 V324 T321 hbond F295 1 SheetConstraint E301 L305 D350 V346 hbond E302 1 SheetConstraint M299 L305 R314 T308 hbond F303 1 SheetConstraint R312 P316 E335 V331 hbond Y313 1 SheetConstraint G343 F348 D361 G356 hbond V346 1 SheetConstraint F348 D350 L355 E353 hbond D350 1 # A290-I373/decoys/try1-opt2.sheets: SheetConstraint A290 V293 E302 M299 hbond A290 1 SheetConstraint V293 F295 P300 E298 hbond F295 1 SheetConstraint M299 E306 I373 D366 hbond M299 1 SheetConstraint Y313 Q315 L355 E353 hbond R314 1 SheetConstraint K319 V324 Y349 Q344 hbond V320 1 SheetConstraint V320 G326 I373 G367 hbond T321 1 SheetConstraint K325 E327 V369 G367 hbond G326 1 SheetConstraint K328 I332 T341 Q337 hbond T329 1 SheetConstraint I332 A334 L355 E353 hbond I332 1 # L171-P300/decoys/try1-opt2.sheets: SheetConstraint M223 V225 M230 R228 hbond V225 1 SheetConstraint L236 M237 K265 G264 hbond M237 1 SheetConstraint M237 G242 V263 A258 hbond Y239 1 SheetConstraint N270 V274 R297 V293 hbond I271 1 SheetConstraint G275 Q276 V293 Q292 hbond Q276 1 SheetConstraint G275 Y279 Q292 L288 hbond F278 1 # M1-A183/decoys/try1-opt2.sheets: SheetConstraint R8 M13 D32 F37 hbond V9 1 SheetConstraint R8 V11 D122 A125 hbond V10 1 SheetConstraint G35 N40 Y70 F75 hbond G35 1 SheetConstraint Y123 T126 K166 F169 hbond V124 1 # M1-A183/decoys/try2-opt2.sheets: SheetConstraint T7 G12 Y31 I36 hbond V9 1 SheetConstraint V9 V10 Y123 V124 hbond V10 1 SheetConstraint I34 K39 P69 N74 hbond G35 1 # M1-K205/decoys/try1-opt2.sheets: SheetConstraint R8 M13 D32 F37 hbond V9 1 SheetConstraint V9 V11 Y123 A125 hbond V10 1 SheetConstraint G35 N40 Y70 F75 hbond G35 1 SheetConstraint Y123 T126 K166 F169 hbond V124 1 # M1-K205/decoys/try2-opt2.sheets: SheetConstraint I34 F37 P69 S72 hbond G35 1 SheetConstraint D122 H128 L188 K194 hbond V124 1 # M1-W260+F311-I373/decoys/try1-opt2.sheets: SheetConstraint T7 M13 Y31 F37 hbond V9 1 SheetConstraint I34 N40 P69 F75 hbond G35 1 SheetConstraint D122 Y123 R8 V9 hbond Y123 1 SheetConstraint Y123 T126 K166 F169 hbond V124 1 SheetConstraint D195 C200 K208 G203 hbond S196 1 SheetConstraint D226 D229 I249 Q252 hbond R228 1 SheetConstraint V296 Y299 L305 E302 hbond V297 1 I've put these constraints into "collected.constraints" and also into try11.costfcn. They are almost certainly inconsistent constraints, but we can see which models do the best job of matching them. Currently try8-opt2 does the best, with try10-opt2 close behind. I'll start a try11 run which uses these constraints and the alignments from both the master directory and the subdirectories (though not the two new ones that I just started). try11 started on peep. Sun Jun 25 10:56:28 PDT 2006 Kevin Karplus I just noticed that the try9 and try10 costfcns have T0289 in them (rather than T0316) so the pred_alpha components are ignored. I also can't use the M1-W260+F311-I373 constraints, because the numbering is not consistent. Fixing these problems still left try8-opt2 at the top, with try10-opt2 second. try11 restarted on peep. Sun Jun 25 17:08:11 PDT 2006 Kevin Karplus try11 scores almost as well as try8 with the try11 costfcn (an repacking improves it), but it does not do quite as well with the unconstrained costfcn (after try8, try10, try6m try9, try4). Sun Jun 25 18:44:43 PDT 2006 Grant Thiltgen I started try12. It's another chimera with try2 of M1-A183 and try6 of A130-I373 cut at residue 131. I used protein shop to alleviate some of the clashes, and I'm hoping undertaker doesn't tear it up too much. Since I messed up the pred_alpha components with try9 and try10, I started try13 which is the polishing run of the rosetta models and try 14 which is the polishing run of try8 and try6, both of which have the correct pred_alpha in them. Mon Jun 26 08:10:23 PDT 2006 Kevin Karplus try15.costfcn has the try11.costfcn constraints plus # M1-D137/decoys/try1-opt2.sheets SheetConstraint (T0316)R8 (T0316)M13 (T0316)D32 (T0316)F37 hbond (T0316)V9 1 SheetConstraint (T0316)V9 (T0316)V10 (T0316)Y123 (T0316)V124 hbond (T0316)V10 1 SheetConstraint (T0316)I34 (T0316)N40 (T0316)P69 (T0316)F75 hbond (T0316)G35 1 # D122-K269/decoys/try1-opt2.sheets SheetConstraint (T0316)H141 (T0316)V146 (T0316)P170 (T0316)E175 hbond (T0316)M142 1 SheetConstraint (T0316)S189 (T0316)K193 (T0316)I202 (T0316)G198 hbond (T0316)T190 1 SheetConstraint (T0316)Y238 (T0316)I241 (T0316)K265 (T0316)S268 hbond (T0316)Y239 1 It favors try12, try8, try14, try11, try10, try6, ... On second thought, I think I should make all constraints in try15 be bonus constraints, since I *know* that they are inconsistent. That changes the order to try14, try10, try8, try6, try12, try9, try13 Obviously, none of these models is right, but which is the least wrong? I think we may need to look for subdomain models that are good and put them together, even if we can't close the gaps between them. Mon Jun 26 8:22:16 PDT 2006 Grant Thiltgen Well, try14 and try10 are polished models of try8, and try13 is a polished model of try9, so they are similar. In that case, I'd probably use try14, try6, try12, and try9, and figure out which may be better from those. It still leaves us one model to submit though. Mon Jun 26 11:23:58 PDT 2006 Kevin Karplus I modified try15 again, this time reducing clashes and breaks, since they are dominating now that the constraints are all bonus constraints. Order is changed again: try8, try14, try12, try10, try6, try9, try13, try2, try1 Grant's order looks reasonable still, since among similar models the fewer clashes and breaks the better. I have started try15 (from alignments again) on whidbey. I think we could have done a better job on this model. Mon Jun 26 12:15:23 PDT 2006 Grant Thiltgen The big problem I've run into with this protein is that most of the tRNA methyltransferases are around 210-250 amino acids in length and this one is 373 amino acids. There's a lot of extra stuff in it. I found some tRNA formyltransferases in the PDB, one of which is in the str2 fold recognition set. 1fmtA is in the set with an e-value of 1.24e-02. The other formyltransferase I found is 2fmtC. Both of these are 325 proteins. I started try16 on vashon using those two proteins for alignments. I used the same costfcn as try15, with the bonuses for the sheets. I also started try17 on orcas using those two proteins and the same costfcn as try7. Mon Jun 26 16:51:02 PDT 2006 Grant Thiltgen I looked at the different regions for piecing together a mix of good alignments. I liked the first run for D122-K269, but I can't find a good place to cut it for adding the barrel in the C-terminal domain. I started try2 of this region to see if there's anything better while I'm trying to see if I can piece together what is already there. I've used try2-opt2 from M1-A183 and try1-opt2 from D122-K269 to make the first two parts. I cut the first half at residue 131. For the next half, I used try4-opt2 from A130-I373, and I cut it at residue 267. I started try18 on orcas with the chimera to see if undertaker can remove some of the clashes and move things around. try19 is when I manually tried to remove some of the clashes with proteinshop and then ran it through undertaker to see what it can do. Mon Jun 26 18:14:41 PDT 2006 Grant Thiltgen Currently, the best results from the unconstrained costfcn are the following: try8, try14, try10, try12, try6, try9, try13, try11 The best results from the try15 costfcn: try8, try14, try12, try10, try6, try9, try13, try15 For submissions, I would use: 1) either try8 or try14 (try14 is a polished version, but seems to score slightly less) 2) try12 3) try6 4) try9 5) try11 or try15 or 1) either try8 or try14 (try14 is a polished version, but seems to score slightly less) 2) try12 3) try9 4) try11 5) try15 Since I have more trys running, I will check those tonight and find out how well they work in comparison to these current top 5. Mon Jun 26 20:36:56 PDT 2006 Kevin Karplus I submitted try8-opt2 try14-opt2 (which came from try6, not try8) try9-opt2.repack-nonPC try11-opt2 try15-opt2 I don't really like try15-opt2, and would be glad to replace it with a better alternative. Tue Jun 27 00:56:30 PDT 2006 Grant Thiltgen try18 and try19 are finally finished. try19 scores behind try10 and try12, but better than try9, try11 and try15. try18 follows try19, and also scores better than try9, try11, and try15. I started try20 on shaw which is attempting to refine try18, and try21 on lopez, which is attempting to refine try19. I would suggest removing try15 and putting in try19 or try18 at this point. Or maybe even try11 and try15 and adding try19 and try18. It's a bit of a toss up though. Tue Jun 27 07:38:15 PDT 2006 Kevin Karplus I like try21 better than try20. Rosetta does too, putting it behind just try10 and try9. I scored it with the try15 costfcn (which has the large set of inconsistent bonus constraints) and try11.costfcn (which has a somewhat smaller set of inconsistent constraints). Try21 scores just behind try8 on the try15 costfcn and is at teh top for the try11 costfcn. I'll put try21 on the top and push all the rest down one, dropping try15 off the bottom. Tue Jun 27 07:51:46 PDT 2006 Kevin Karplus Resubmitted: ReadConformPDB T0316.try21-opt2.pdb ReadConformPDB T0316.try8-opt2.pdb ReadConformPDB T0316.try14-opt2.pdb ReadConformPDB T0316.try9-opt2.repack-nonPC.pdb # from ROBETTA_TS4 ReadConformPDB T0316.try11-opt2.pdb I had a hard time trying to figure out what the history of the models was. There was not enough information in the README file to figure out what was done in creating the chimeras (which parts were taken from which previous models), and there is no log trail elsewhere to reconstruct it from. We need more info in the README files! These are our lab notebooks, and need to be complete!