Citat af John Kuszewski <johnk at mail.nih.gov>:
Please see below for comments/questions. > Hi Jakob, > > On Aug 15, 2007, at 8:39 AM, Jakob Toudahl Nielsen wrote: > >> Dear Xplor-nih users, >> >> I am using PASD (which I think is a very elegant method) to calculate >> the structure of a protein with known structure using solid state NMR >> data. I have derived the peak assignments with our new method which >> uses statistical data and secondary designition based on chemical >> shifts. For each individual peak assignment within each restraint I >> have also derived an assignment likelihood for each assignment. > > On what basis did you assign the peakAssignments' likelihoods? I use statistics performed on several hundred protein structure corelating structure and secondary designation built into our program to estimate the likelihoods. > >> Since my use of PASD is somewhat different from normal use, I would >> like to play around with a few parameters, e.g. a longer duration of >> the first heating phase in pass2 etc... Is this possible? what would >> be the syntax? > > Of course it's possible. The pass2 protocol is defined in > xplor/tcl/sa_protocols.tcl > Search through it for a TCL procedure called 'pass2'. Within it, > you'll see calls > to TCL procs called highTemp and cooling, with lots of flags to each > call, defining > virtually everything of interest. > > If you run into something you can't change easily (or don't understand), just > email me. > >> I notice there is a flag "-numStructs" for the pass2 >> procure, but I am not aware of other possible flags/options - is this >> true? > > The best way to proceed would be to make a copy of the pass2 TCL proc, > give it a new name (like 'myPass2'), and then start making changes within > the new proc. > >> So far my results are disapointing, the structures I derive are mostly >> elongated, despite that there a several restraints with long-range >> assignments with significant likelihood between the two ends of the >> chain in a parallel beta-sheet. I also notice that only a few >> (typicallly 50 out of 266) restraints are active at the end of pass2 - >> is this normal/good? > > You only have 266 restraints? How big is your system? Marvin is generally > more successful with more complete datasets. My system is 76 residues. I count the restraints differently, I combine the restraints on both of the diagonal into one single restraint and, furthermore, filter out the restraints with no information (i.e. one and two-bond restraints). Finally my program removes restraints with very small likelihoods. Having said this, I have rather few restraints - we had hoped for better sensitivity... > > Since you have a reference structure, are you using it in the initialMatch > and summarize_pass scripts? That'll give you some valuable statistics in the > headers of your .peaks and .shiftAssignments files. I don't use these scripts. >> What is the violation energy for an inactive >> restraint - is it 0.0?. > > Yes. A peak is inactive if all of its peakAssignments are inactive. > Inactive peakAssignments > have zero energy and force. > >> I was wondering if the structure could be >> trapped in an elongated structure with most of the restraints >> inactivated, do you think this is possible? > > It's possible. Especially if you don't have too many long-range peaks > per residue. I have 16 long-range restraints (i.e. restraints with solely long-range assignment posibilities) of which 13 agree with the structure. I think this is few. > > Given your unusual situation, you might have more success starting your > calculations > with the pass3 protocol, which uses a quadratic potential instead of > the linear potential. > It's much easier for simulated annealing to optimize. The drawback is > that it's not quite > as tolerant of very large numbers of bad peaks, but it can still do a > good job on its own. > If you want to use the pass3 protocol for starters, you can and should > still repeat it--ie., > run a pass3 protocol, summarize it, run another pass3 (with different > filenames for the PDBs, > of course!), summarize it, and so forth. > > Hope this helps. > > --JK Many thanks for your help, I aggreed with you and started from pass3 which I tried to modify in different ways. However, I still have problems getting my system to converge. Starting from pass3 still gives mostly elongated, featureless structures satisfying none or very few of the long-range restraints. I increased the noe force constant, kNOE, 10-fold. Analyzing the log-file showed that the restrain violation energy increased 10-fold accordingly but no structures converged. I realize that pasd protocols differ significantly from other annealing protocols in that the noe constant is ramped up during cooling rather ramped down. I guess the procedure must be followed in order to ensure no bias towards prior likelihoods. I tried a protocol in which the kNOE is ramped up during annealing - with no succes. I another run a used the pass3 protocol again with 10-fold increased kNOE but with ramping up the kNOE, but this time I used only the 16 long-range restraints. Surprisingly, about half of the structures converged with near-zero violation violation energy. I analyzed the log-file and noted that the violation energy reported by Xplor for the non-converged structures were about 10-fold higher than the previous with exactly same parameters but with all 266 restraints. I have attached a diagram showing the violation energy during annealing for the system using only the long-range restraints (broken lines) . Is it true that pasd somehow scales the violation energy to agree with the total number of NOEs? Does it have anything to do with the concepts of "-noeCompletenessWeight" or "-scatterWeight" in the sa_protocols.tcl script - I don't understand the meaning of these variables. Obveously, I would like my calculations to converge using all NOEs, have you had similar problems? I notice that you choose the number of long-range violations as your criteria for convergence rather than the total energy. Is this because there were to many structures with lower total energy being incorrectly folded? Do you have suggestion of what goes wrong in my case? Since I use solid state data I have restraints between carbon atoms. The ambiguity is rather large for my restraint ca. 4-5 degeneracy (as a geometric average) after doing prefiltering. I think this should not be a problem. I hope you can help me. Best regards, Jakob Nielsen ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.
