Thu May 11 08:40:39 PDT 2006 T0284 Make started Thu May 11 08:41:36 PDT 2006 Running on cheep.cse.ucsc.edu Thu May 11 08:57:24 PDT 2006 Kevin Karplus Since the T06 alignment finds 14 structures in PDB, this looks like it is a comparative modeling problem. 1oqf[AB] = 1mum[AB] 1ujq[ABCD] = 1o5q[ABCD] 1xg4[ABCD]=1xg3[ABCD] 1igw[ABCD] 1f8m[ABCD]=1f61[AB] 1f8i[ABCD] 1dquA 1zlp[AB] 1m3u[ABCDEFGHIJ] 1s2wA=1s2v[ABCD]=1s2t[AB]=1m1b[AB] 1s2u[AB] 1oy0[ABCDE] 1pym[AB] 1o68[ABCDE]=1o66[ABCDE] We may need to do multimeric modeling to do a good job on this protein. The t2k alignment finds 12 different PDB sequences, and t04 finds all 14. The t04 and t06 alignments have very similar numbers of sequences (1213 and 1210, respectively). Thu May 11 10:56:23 PDT 2006 Kevin Karplus The hits are to TIM barrels: Superfamily: Phosphoenolpyruvate/pyruvate domain, mostly Family: Phosphoenolpyruvate mutase/Isocitrate lyase-like. Note: this family forms swapped dimers! Make started Thu May 11 11:47:21 PDT 2006 Running on cheep.cse.ucsc.edu We will almost certainly have to build dimers of this target and we need to choose good templates. I ran the following commmands to find out how good a match the templates were: foreach x ( */*simplesw*dist.gz ) echo -n $x " " >> simple-scores gunzip -c $x | grep -v '%' | grep -v T0284 >> simple-scores end The top few hits were 1mumA, 1xg4A, 1ujqA, 1zlpA, 1pymA, 1s2wA, 1s2uA All the hits after that were very much poorer scores with the simple model. Two of these (1zlpA and 1s2uA) aren't in the template library, but the rest are. We probably want to align to and copy the dimer from 1mum[AB], as the most similar sequence. We'll probably have to add some constraints to model the missing Mg ion, which in 1mumA is coordinated by D58, D85, D87, E115 and R158 (though R158 is a bit separated from the rest). In the target, these residues correspond to D61, D88, D80, E117, and R159. It would be a good idea for this model to modify undertaker's SCWRL call for initial alignments to preserve the rotamers for identical residues. Then we could pick up constraints from the model fairly easily, without having to measure them and copy them from 1mumA (and the critical residues are much more likely to be in the correct rotamer). According to BLAST at NCBI, the best PDB hits are 1oqfB = 1oqfA = 1mumA 1xg4D = 1xg4A 1ujqD = 1ujqA 1zlpB = 1zlpA 1s2wA 1pymB = 1pymA 1s2uB = 1s2uA These are pretty much in agreement with the SAM hits. Interestingly, the longest BLAST alignment (at the C-terminus) is the one to 1zlpA, which has a different dimerization than the other dimers, with the monomers essentially independent. We may want to model the C-terminus off of 1zlpA, particularly if we submit a monomeric solution. Fri May 12 11:03:18 PDT 2006 Kevin Karplus For try2, I set up constraints from the try1-opt2.sheets and try1-opt2.helices and (weaker) the rr constraints. Fri May 12 11:43:39 PDT 2006 Kevin Karplus The try2 run seems to be basing its predictions on the 1xg4A alignments, particularly 1xg4A-T0284-t2k-local-adpstyle5.a2m. This is the same template that try1-opt2 favored, even though the sequence aligners thought that 1mumA was better. Fri May 12 11:59:56 PDT 2006 Kevin Karplus I went to pqs (the Protein Quaternary Structure database) to look at the multimeric structure for 1zlp. It turns out it dimerizes just the same as the rest---the crystallographers just picked a goofy unit cell, with more contacts between copies than within the cell. Here are som constraints from 1mumA: Distance ASP58A.CG-ASP85A.CG: 7.286 Distance ASP87A.CG-ASP58A.CG: 6.168 Distance GLU115A.CD-ASP58A.CG: 8.386 Distance ASP87A.CG-ASP85A.CG: 4.423 Distance ASP85A.CG-GLU115A.CD: 6.388 Distance GLU115A.CD-ASP87A.CG: 5.098 Distance ARG158A.NH1-ASP58A.CG: 8.003 Distance ARG158A.NH1-ASP85A.CG: 5.463 Distance ARG158A.NH1-ASP87A.CG: 7.079 Distance ARG158A.NH1-GLU115A.CD: 4.513 From 1mumB Distance ASP58B.CG-ASP85B.CG: 7.325 Distance ASP58B.CG-ASP87B.CG: 6.234 Distance ASP58B.CG-GLU115B.CD: 8.443 Distance ASP85B.CG-ASP87B.CG: 4.376 Distance ASP85B.CG-GLU115B.CD: 6.294 Distance GLU115B.CD-ASP87B.CG: 5.110 Distance ARG158B.NH2-ASP85B.CG: 7.148 Distance ARG158B.NH2-ASP58B.CG: 7.819 Distance ARG158B.NH2-ASP87B.CG: 7.584 Distance ARG158B.NH2-GLU115B.CD: 4.159 Distance ARG158B.NH1-ASP58B.CG: 7.887 Distance ARG158B.NH1-ASP85B.CG: 5.392 Distance ARG158B.NH1-ASP87B.CG: 7.040 Distance ARG158B.NH1-GLU115B.CD: 4.517 (I really should look at some of the other templates also---perhaps 1xg4A, which undertaker likes so well: T0284 1mumA 1xg4A D61 D58 D58 D88 D85 D85 D90 D87 D87 E117 E115 E115 R159 R158 R158 From 1xg4A: Distance ASP58A.CG-ASP85A.CG: 7.137 Distance ASP58A.CG-ASP87A.CG: 6.263 Distance ASP58A.CG-GLU115A.CD: 8.133 Distance ASP58A.CG-ARG158A.NH1: 7.565 Distance ASP85A.CG-ASP87A.CG: 4.684 Distance ASP85A.CG-GLU115A.CD: 6.336 Distance ASP85A.CG-ARG158A.NH1: 5.623 Distance ASP87A.CG-GLU115A.CD: 5.188 Distance ASP87A.CG-ARG158A.NH1: 7.457 Distance GLU115A.CD-ARG158A.NH1: 4.614 From 1xg4D: Distance ASP58D.CG-ASP85D.CG: 7.114 Distance ASP58D.CG-ASP87D.CG: 6.303 Distance ASP58D.CG-GLU115D.CD: 8.107 Distance ASP58D.CG-ARG158D.NH1: 7.532 Distance ASP85D.CG-ASP87D.CG: 4.736 Distance ASP85D.CG-GLU115D.CD: 6.293 Distance ASP85D.CG-ARG158D.NH1: 5.572 Distance ASP87D.CG-GLU115D.CD: 5.271 Distance ASP87D.CG-ARG158D.NH1: 7.501 Distance GLU115D.CD-ARG158D.NH1: 4.516 From 1xg4B: Distance ASP58B.CG-ASP85B.CG: 7.163 Distance ASP58B.CG-ASP87B.CG: 6.332 Distance ASP58B.CG-GLU115B.CD: 8.177 Distance ASP58B.CG-ARG158B.NH1: 7.632 Distance ASP85B.CG-ASP87B.CG: 4.801 Distance ASP85B.CG-GLU115B.CD: 6.149 Distance ASP85B.CG-ARG158B.NH1: 5.694 Distance ASP87B.CG-GLU115B.CD: 5.150 Distance ASP87B.CG-ARG158B.NH1: 7.659 Distance GLU115B.CD-ARG158B.NH1: 4.649 (1xg4C seems to be missing the Mg ion) Fri May 12 12:42:56 PDT 2006 Kevin Karplus I can make constraints of these: constraint D61.CG D88.CG 7.0 7.2 7.5 constraint D61.CG D90.CG 5.9 6.3 6.5 constraint D61.CG E117.CD 7.9 8.2 8.6 constraint D61.CG R159.NH1 7.3 7.8 8.3 constraint D88.CG D90.CG 4.2 4.6 5.0 constraint D88.CG E117.CD 6.1 6.3 6.6 constraint D88.CG R159.NH1 5.2 5.6 5.8 constraint D90.CG E117.CD 4.9 5.2 5.4 constraint D90.CG R159.NH1 6.9 7.4 7.9 constraint E117.CD R159.NH1 4.3 4.6 4.8 I'll add these contraints to try3.costfcn. Sat May 13 08:23:24 PDT 2006 Kevin Karplus There is actually some ambiguity about the proper alignment of the 7th strand (P206-T211). There are two possibilities 22 CYHTAsvf 43 ECGILgg 84 PVIADad 110 giaALTIEd 154 paLTIIArt 184 GICLVgv 206 pLMLVTyg 229 rVVVng 22 CYHTAsvf 22 CYHTAsvf 43 ECGILgg 84 PVIADad 110 giaALTIEd 152 paLTIIArt 184 GICLVgv 206 pLMLVTyg 229 rVVVng 22 CYHTAsvf I think I like the first of these better, which aligns R229 with L206, as the hydrophobicity match for the barrel edges is better. I will do a monomer run with try3, then try to figure out how to make the dimer. Sat May 13 08:59:22 PDT 2006 Kevin Karplus RATS! undertaker crashed on try3: undertaker: ../ultimate/src/Transform/torsion_angle.h:95: double torsion_angle(const XYZpoint&, const XYZpoint&, const XYZpoint&, const XYZpoint&): Assertion `BA != XYZpoint::ZeroPoint' failed. Now I have to recompile it for the debugger, and run it again with the same seed for the random number generator, and hope that I can figure out where the problem is. Sat May 13 09:17:07 PDT 2006 Kevin Karplus When I tried running under the debugger, the program did not crash. I'll have to try running the optimized version of the code under the debugger, which is not as easy, but should at least tickle the same bug. Sat May 13 10:13:24 PDT 2006 Kevin Karplus I found one bug in Conformation::has_alpha(), but fixing it did not remove the problem. Recompiling just AnglePredCostFunction.cc with DEBUG=1 did cause the problem to go away. Sat May 13 10:36:10 PDT 2006 Kevin Karplus Recompiling with -O3 caused the error to come back. Sat May 13 13:18:42 PDT 2006 Kevin Karplus The bug was *not* a compiler bug as I feared, but 3 colinear points passed to torsion_angle. Adding a colinearity check to has_alpha eliminated the problem. Now I need to add colinearity checks before other torsion angle computations---perhaps for phi and psi. Sat May 13 15:18:12 PDT 2006 Kevin Karplus After running try3, which I expect to do the best job so far on the TIM barrel, I'll try making a chimera between try3-opt2 and the alignment to 1zlpA, which I expect to have the best version of the swapped dimer part. I'll take this monomer and make a dimer of it, then try reoptimizing the dimer. Sat May 13 15:41:09 PDT 2006 Kevin Karplus I made decoys/chimera-try3-1zlpA from try3-opt2 and the 5th model of T0283.undertaker-align.pdb Use T0284.try3-opt2.pdb through residue A236 Use alignment to 1zlpA (from T0283.undertaker-align.pdb model 5) for A237-V285 (and try3-opt2 again for rest). Residues S258 and D259 from alignment to 1xg4A, since they were missing in alignment to 1zlpA. Sat May 13 15:57:48 PDT 2006 Kevin Karplus Aside from the little bit of the c-terminus that was not copiable, the model in dimer/decoys/dimer-1mumA-from-chimera-try3-1zlpA.pdb does not look too bad. There are a few clashes between the monomers, but we may be able to fix that by optimization in the dimer. Mon May 15 08:05:17 PDT 2006 Kevin Karplus After spending most of the weekend chasing down bugs in undertaker, I finally got the dimer/try1 run to complete. Unfortunately, it moved the C-terminal helices so that the two monomers didn't interact, rather than keeping them locked together as desired. I'll have to add some inter-chain constraints, which may be a bit messy. One possible constraint is between A244.CB and its counterpart, keeping them about 3.3 Angstroms between centers. Another possibility is I242.CG2 and T270.CG2 (around 4.8) between monomers. Another constraint within each monomer is an Hbond between R35.NH2 and E274.O Wed May 17 17:24:55 PDT 2006 Kevin Karplus With the computers being down yesterday and today, I've not had much chance to look at or fix the dimer. The try2-opt2 run did not tear the two monomers apart, but I think I have the wrong constraints between the monomers, as there was some breaking of the chain to accomodate them. I'll have to take another look and try out some better inter-monomer constraints. Thu May 18 22:39:05 PDT 2006 Kevin Karplus There are no logs for the dimer try2 run, because I had to run it under the debugger. I *hope* I have now fixed the bugs in undertaker. Here are some extra constraints to try to keep the monomers from drifting apart I46.CA separation 27.86 T25.CA separation 36.05 I86.CA separation 34.19 F239.CA to S266.CA in other chain 5.2 Thu May 18 23:48:10 PDT 2006 Kevin Karplus RATS! the undertaker try3 run crashed, with the same symptoms as before. Just after # Trying to read SCWRLed conformation from /var/tmp/from_scwrl_341772046.pdb we get a rather uninformative assertion failure: undertaker: XYZpoint.h:57: void XYZpoint::unitize(): Assertion `mag>0' failed. I suspect that this is coming from the "symmetrize" operation on the return from scwrl. Even before the crash there was clearly something wrong with symmetrize, as the symmetrized initial conformations (which should be a no-op) rasied the cost enormously. Sat May 20 08:55:30 PDT 2006 Kevin Karplus I got undertaker fixed last night, and the dimer/try4 run worked fine, reducing breaks and contraints without blowing the dimer apart. For try5, I'll keep the Mg coordinating contraints, but throw out the constraints on the dimer, and just polish the existing models. Sat May 20 12:18:28 PDT 2006 Kevin Karplus The try5 run, while not yet completed, does not seem to be doing much---neither the clashes nor the breaks seem to have changed much, though the cost has dropped a little. The try1-opt1 model has no change in the break cost function, and only a small drop in soft_clashes. There are slight decreases in sidechain, dry5, and phobic_fit, and the Mg-coordinating residues fit their constraints slightly better. It may be worthwhile to add some more constraints in the neighborhood of the MG ion, but reduce the constraint weight. I think that maybe N161 needs some constraints on its sidechain, and maybe D118 and Y93, since they are conserved and near the Mg binding site. Sat May 20 15:25:23 PDT 2006 Kevin Karplus The dimer try5-opt2 run looks pretty good, but I think that there is a little tweaking needed for G254-S266 since V256 and L260 ought to be buried against the other monomer, and R253 should be exposed more. I also wonder about the exposure of W278--the final helix may need a quarter turn. Sat May 27 20:57:33 PDT 2006 Firas Khatib wow... Proteinshop and dimers DO NOT mix well! It essentially connects the two and makes a horrible mess! I just saved the monomer (res 1-287) and worked on that. The only way to rotate the helix with R253 was to make A255 into coil (otherwise the helix gets caught on the coil and cannot rotate) I rotated it to expose R253 more and try to bury V256 (which is in a coil, and hard to control in Proteinshop) I then rotated the small helix with L260, but couldn't turn it too much and rotated the helix containing W278 a bit as well. I saved this as: try5-opt2.chain1.rotate1.renum.pdb in casp7/T0284/dimer/decoys/Proteinshop Sat May 27 22:23:25 PDT 2006 Firas Khatib I tried Proteinshoping the monomer with the dimer loaded, which makes it extremely difficult the closer you are to the C-terminal end of the chain (and the N-terminal end of the other monomer since Proteinshop connects them). in casp7/T0284/dimer/decoys/Proteinshop I saved 3 pdb files that go from least tweaking: try5-opt2.proto.dimerTake1.pdb to more drastic tweaking: try5-opt2.proto.dimerTake2.pdb to most tweaking: try5-opt2.proto.dimerTake3.pdb I would try sending all 3 to undertaker and if it can deal with Take3 or 2 that's great, but otherwise it should be able to handle Take1 all right. Kevin, can you run them as try5, try6, and try7? I have never inputted a dimer and don't want to mess anything up. Sun May 28 08:57:16 PDT 2006 Kevin Karplus A tweaked monomer can be made into a dimer by the method described in casp7/README and the example in casp7/T0284/make-dimer.under I'm going to move the three tweaked dimers up from decoys/Proteinshop/ to decoys/ and gzip them. I'll then try simultaneously optimizing them as try6 in dimers (there is already a try5 in dimers). Sun May 28 13:21:53 PDT 2006 Kevin Karplus try6-opt2 never got as good a score as try5-opt2. I don't really see very much difference between try4-opt2, try5-opt2, and try6-opt2. R253 is slightly more exposed and V256 a little more buried in try6-opt2, but L260 is essentially unchanged. The later helices are essentially unchanged. Sun May 28 13:35:40 PDT 2006 Kevin Karplus None of the protoshop runs Firas mentioned seem to have rotated the final (A277-M283) helix. Sun May 28 14:32:04 PDT 2006 Firas Khatib I had mentioned: "and rotated the helix containing W278 a bit as well. I saved this as: try5-opt2.chain1.rotate1.renum.pdb in casp7/T0284/dimer/decoys/Proteinshop" this was the only model where I was able to rotate that helix at all (because it was just the monomer... in the context of the dimer that helix cannot be rotated in Proteinshop) so if you look at try5-opt2.chain1.rotate1.renum.pdb, the A277-283 helix is rotated a bit, but you will notice that the coil next to it (at Q276) cannot be bent back any further, so this helix cannot rotate any further without changing either 275 or 277 from helix to coil. Date: Sun, 28 May 2006 14:54:53 -0700 From: "Firas Khatib" To: "Kevin Karplus" Hey Kevin, I saw in the T0284 README that none of the models had the A277-M283 helix rotated. In all the dimerTake models this is because the c-terminal end of the monomer was connected to the n-terminal end of the other, thus any rotation of that last helix was impossible. When I saved try5-opt2 as just a monomer and worked with that I was able to rotate that helix a bit: try5-opt2.chain1.rotate1.renum.pdb but as I just added to the README, I could not rotate that helix any further without changing residues 275 or 277 from helix to coil to give it more slack. Let me know if you want me to try that and I can do it. ---Firas On 5/28/06, Kevin Karplus wrote: > > I started T0284/dimer/try6 using all three of your tweaked dimers as > starting points. It will probably work most with > try5-opt2.proto.dimerTake1, because that costs least (159.514) with > the try6.costfcn, but there may be some crossover. > > If you want to start runs from just Take3 or Take2 and Take3, you can > copy the try6.under file to make try7.under or try8.under and remove > the Take1 reading. > > Kevin Karplus > > ------------------------------------------------------------ Date: Sun, 28 May 2006 15:01:11 -0700 To: "Firas" From: Kevin Karplus Subject: Re: starting to send off targets OK, I'll try making a dimer from try5-opt2.chain1.rotate1.renum.pdb and optimizing that. Date: Sun, 28 May 2006 15:02:37 -0700 From: Kevin Karplus To: "Firas" Subject: Re: starting to send off targets Actually, if you can rotate the helix further (even with a little unwinding of the helix), I can make dimers from both and optimize from there. kevin ------------------------------------------------------------ Sun May 28 15:46:38 PDT 2006 Firas Khatib I took in try5-opt2.chain1.rotate1.renum.pdb and rotated helix A277-M283 more. The c-terminus after the helix is a bit iffy, but hopefully Undertaker will fix that. The result is: try5-opt2.chain1.rotate2.renum.pdb in casp7/T0284/dimer/decoys/Proteinshop this file has R253 exposed pretty well and V256 buried more, but L260 is not buried and that helix will rotate no further (you can see that S258 has nowhere to go except into the large helix if we rotate more) so I will try to manually move that helix so that it is rotatable. For now, I think that optimizing from try5-opt2.chain1.rotate2.renum.pdb is worthwhile and I will work on L260. Sun May 28 16:54:10 PDT 2006 Kevin Karplus I made dimers from try5-opt2.chain1.rotate1.renum.pdb try5-opt2.chain1.rotate2.renum.pdb try5-opt2.chain1.rotate3.renum.pdb calling them rotate1.pdb, rotate2.pdb, and rotate3.pdb. Of them, rotate1 scored the best, due mainly to clash differences. They all had clash problems, but these may be fixable with tiny tweaks to the rotamers and to the dimer transform, so I'll start try7 optimization using the rotate2 dimer only. Sun May 28 17:12:54 PDT 2006 Firas Khatib try5-opt2.chain1.rotate4.renum.pdb might have less clashes than try5-opt2.chain1.rotate3.renum.pdb it might be worth a try... both rotate3 and rotate4 only rotate the L260 helix and are based off of rotate2 Sun May 28 22:44:06 PDT 2006 Kevin Karplus try7-opt2 was a complete waste of computer resources---at some point it tore the dimerization region apart, in frustration over the difficulty of packing it tightly. I made rotate4.pdb, but it has even worse clashes than rotate3.pdb. I hesitate to do any optimization with them, without some constraints to keep the dimerization interface more or less intact. Perhaps I should pick up some constraints from try5-opt2 and loosen them a little so that residues can move. Too tired to do this tonight, but I'll try to look at it tomorrow. Tue May 30 15:58:35 PDT 2006 Firas Khatib I'm going to work off of try5-opt2 once more, using just the monomer and rotating each of the 3 helices on their own, saving them to three files. quick question, you mentioned on sunday: "Of them, rotate1 scored the best, due mainly to clash differences. They all had clash problems, but these may be fixable with tiny tweaks to the rotamers and to the dimer transform, so I'll start try7 optimization using the rotate2 dimer only." Did you try using the rotate1 dimer only since it scored the best? I am running try8 with rotate1 as the dimer. Tue May 30 17:43:34 PDT 2006 Firas Khatib quick question: I was looking at T0284.try5-opt2.pdb using the "near" script for burial (which is what I have been using for CASP until now) but then I looked at it with the "burial" script (that Grant had been using) and looked at: L260 in both versions (since we want it buried against the other monomer) Using "near" (which is from T04-near-backbone-11, right?) it is green (which means it could be exposed or buried) But using "burial" (which is from T04-CB_burial-14-7, right?) it is cyan (which is more exposed than buried, right?) Is it just that cyan isn't even blue (so it's more "maybe" than "exposed") or should L260 not really be buried? None of them really have any strong burial predictions for L260 in summary.html except: T0284.t2k.CB_burial_14_7-logo.pdf It seems to predict "A/B" the most for L260. I guess my main question is: which script do you trust the most when you are looking at burial? Tue May 30 19:37:40 PDT 2006 Kevin Karplus dimer/try6 was run with try5-opt2.proto.dimerTake1.pdb try5-opt2.proto.dimerTake2.pdb try5-opt2.proto.dimerTake3.pdb as starting points dimer/try7 was run with rotate2.pdb Firas is running dimer/try8 with rotate1.pdb Try8-opt1 has destroyed the dimerization interface. I think we need to add some constraints (like in try4.costfcn, but perhaps more carefully chosen) to keep the interface from getting shredded. I'll do that for try9, but without taking the time to get the constraints right. I'm not sure what script I believe most for burial. Two years ago, the near script was the most useful, but we are using different networks now (and still haven't retrained any networks for higher accuracy), so the secondary predictions may not be as good as they should be. In any case, I would expect a fairly high probability for burying a leucine---they are very hydrophobic. Tue May 30 20:54:23 PDT 2006 Firas Khatib indeed, that's why I was surprised that T0284.t2k.CB_burial_14_7-logo.pdf predicted a Leucine to be exposed! I'll stick with the "near" script then... Wed May 31 11:15:54 PDT 2006 Kevin Karplus I have modified superimpose-best.under to pick up the momomers from the unpacked versions of the three best dimers (try5-opt2, try6-opt2, and try8-opt2). Inicidentally, Firas, can ProteinShop handle dimers when they are properly in separate chains like in the T0284.try5-opt2.unpack.pdb.gz file? try8 is wrong---I'll remove it. try9 failed with an assertion failure in AlignedFragments---the assertion that checks that edge-length of Tertiary edges is correct. I don't think I have time to debug this now. Wed May 31 11:39:00 PDT 2006 Kevin Karplus I emailed dimer/try5-opt2 dimer/try6-opt2 undertaker-align model 1 from 1s2wA undertaker-align model 2 from 1xg4A undertaker-align model 3 from 1o66A I first copied the generic method file from the starter directory and edited it, adding a METHOD_FILE:= line to the Makefile for the local copy. After "make casp_models" I edited model3.ts, model4.ts, and model5.ts to fix the PARENT remark. Then I did "make email" to send the 5 models. Wed May 31 17:12:43 PDT 2006 Firas Khatib >Firas, can ProteinShop handle dimers >when they are properly in separate chains like in the >T0284.try5-opt2.unpack.pdb.gz file? sadly, I haven't been able to get this to work (not just because shaw keeps crashing). the number of chains shown can be modified in Proteinshop.cfg To load all the chains you just have to comment out one line: # chainSelection But when it loads them all in it connects the chains for some reason! This is also true for pdb files with multiple chains from the PDB. My only hope is that they fixed this in Greg's version of Proteinshop. I am still using: /projects/paro/src/ProteinShop2.x-bin-Linux9.0 until Greg gets his version fixed. I've tried using /cse/grads/gregd/ProteinShop/bin but cannot seem to load any files without it crashing. I emailed Greg about this. Make started Tue Jul 18 09:49:59 PDT 2006 Running on lopez.cse.ucsc.edu Tue Jul 18 10:28:55 PDT 2006 Kevin Karplus Because of a typo, I accidentally redid the make for this target. This created a number of local structure predictions that were not available at the time we did the submission. The top few templates may have been rearranged as a result, but the basic hits seem to be the same. Tue Jul 18 22:41:56 PDT 2006 Kevin Karplus NOTE: superimpose-best.under was overwritten by the accidental make. It will have to be re-created if best-models.pdb.gz ever needs to reflect the submitted models. Make started Wed Jul 9 21:31:14 PDT 2008 Running on orcas.cse.ucsc.edu