PASD: converting restraints to xplor

Dear Xplor developers,

I have another question regarding a PASD script in Xplor 2.18. After 
everything ran as desired now so far, I wanted to try and see what I get 
out of a structure calculation using marvin restraints converted to 
xplor if I also include peak assignments with likelihoods lower than the 
standard 0.9. So I re-ran convert_restraints_to_xplor.tcl with a lower 
cutoff:

removeLowLikelihoodPeakAssignments \
     -pot $3dc625 \
     -likelihoodCutoff 0.5

but the xplor_noes file I get out is identical to that produced with 
-likelihoodCutoff 0.9. Upon checking quickly, I right away found one 
peak that has two assignment possibilities with likelihoods between 0.5 
and 0.6 and none larger than that (in the final.peaks file), so I guess 
these should show up in the xplor_noes file if likelihoodCutoff 0.5 is 
used - but the peak is still listed with a "no assignments for this 
peak". Have I missed setting some parameter somewhere else?

Thanks a lot in advance, best greetings,
--robert.

Re: Ensemble calculation of denatured protein

Hello Jie-rong--

> 
> Delta i (eq 14 in 2004 Iwahara JACS) was set to a minimum value of 1.0(/s)
> as mentioned in this paper.  However, the delta T2 I obtained is measured by
> conventional N-15 relaxation methods instead of yours (setGammaI was changed
> therein).  The delta T2 and error (from two repeated experiments fitting
> with several different relay times) are both smaller than 1H delta T2.  Any
> place I can change this setting?  Is that the reason I have some "VIOLATED"
> messages? For example,

the value w_i is set by you for each restraint in the restraint table -
the is the value after the PRE rate. Eq. 14 was used in the paper to
choose this value, but you have complete control over this in your
calculations. 

best regards--
Charles

Re: PASD: converting restraints to xplor

Hi Robert,

Try -cutoff instead of -likelihoodCutoff.  That should correct the  
problem.

--JK

On May 5, 2008, at 12:40 PM, Robert Schneider wrote:

> Dear Xplor developers,
>
> I have another question regarding a PASD script in Xplor 2.18. After
> everything ran as desired now so far, I wanted to try and see what  
> I get
> out of a structure calculation using marvin restraints converted to
> xplor if I also include peak assignments with likelihoods lower  
> than the
> standard 0.9. So I re-ran convert_restraints_to_xplor.tcl with a lower
> cutoff:
>
> removeLowLikelihoodPeakAssignments \
>      -pot $3dc625 \
>      -likelihoodCutoff 0.5
>
> but the xplor_noes file I get out is identical to that produced with
> -likelihoodCutoff 0.9. Upon checking quickly, I right away found one
> peak that has two assignment possibilities with likelihoods between  
> 0.5
> and 0.6 and none larger than that (in the final.peaks file), so I  
> guess

Charles <at> Schwieters.org: Re: longer time taken with parallelization

This may be of general interest:

Hello Nah--

> I am noticing a big performance issue when scaling up, albeit the fact
> that these just trivially parallel calculations with no inter-process
> communications.  For ex., even with my PC, i can calculate say 2
> structures in 5 mts (same script with num_Strucures = 2) when trying
> the parallel version of it with 64 processors, 1000 structures
> calculation took 2.30 hrs (whereas it should have taken ~1 hr from an
> approximation of ~4mts per structure calculation) and interestingly
> with 96 processors and for 2000 structures it's taking close to 9hrs
> for the calculation !!!  /cluster/softs/xplor-nih-2.19/bin/xplor -py
> -parallel -machines 96_procs -o mod.out mod_anneal1.py and the machine
> file 96_procs has 96 processors listed (since each machine has 8
> processors, i am repeating the machine name 8 times ; 12*8 =96

Re: longer time taken with parallelization

_______________________________________________
Xplor-nih mailing list
Xplor-nih <at> nmr.cit.nih.gov
http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih

Re: longer time taken with parallelization

Hello Nah--

>    Though I was aware of the non-linearity in scaling ( i wasn't expecting
>    8X speed in processing), why should 8 different xplor processes be not
>    using the 8 different processors fully ( though limited by memory bus)
>    while other programs (MPI as well as non-MPI) do utilize ~100% of each
>    of the cores.

Different applications will display different behavior depending on how
much memory bandwidth is required. Xplor-NIH has large bandwidth
requirements. So, there just isn't enough data getting to the cores to
keep them busy- they're mostly waiting.

>    Anyways, do you suggest doing calculations in batches - like 12 or 24
>    (1 or 2 processors per node) at a time to maximize the resource utility
>    ?

From what you describe, it sounds like 2 jobs per processor, or four per
node should be right. So, if 1 process per node computes a structure in x
secs, 2 processes per node should calculate two structures in x secs,
and 4 processes per node should calculate four structures in x secs. You
might want to check this. I expect that 6 processes per node will reduce
overall throughput.

>    I'm curious about the 64 vs. 96 processor results: In one case you ran
>    on 8 processors and the other on 12? Or was there some other
>    difference?
>    Yeah, you are right. In one case it was on 8 nodes (8*8 = 64
>    processors) and other in 12 of them. But still, the difference was so
>    much so that cannot be explained simply by non-linear scaling !

I suspect that something else was going on in this case, like another
jobs running on some nodes. Anyway, please run the experiment to find
the sweet spot for your processors and use that in future calculations.

best regards--
Charles

parallel marvin

Hi,
I'm trying to run Marvin an a 16-dual processor cluster and am getting connect failures.  
I'm using:

xplor -parallel -machines ./machine_name_file -pport 22 -tcl sa_pass2.tcl > sa_pass2.tcl.out

but still getting:

connect attempt #28 to c1:22...
connection fault: (-2, 'Name or service not known')

messages.  I can ssh c1 -p 22 without problem, and namd2 distributes fine.  Is there something else I need to
set for -pport (or -port, for that matter)?

Also,
If I have a reasonable homology model, can I use it to start with?  How do I do this?

Thanks,
David Horita

Re: parallel marvin

Hello David--

> I'm trying to run Marvin an a 16-dual processor cluster and am getting
> connect failures.   
> I'm using:
> 
> xplor -parallel -machines ./machine_name_file -pport 22 -tcl sa_pass2.tcl > sa_pass2.tcl.out
> 

don't use port 22 for this- ssh uses this. Use a number larger than 1024
for this.

> but still getting:
> 
> connect attempt #28 to c1:22...
> connection fault: (-2, 'Name or service not known')
> 
> messages.  I can ssh c1 -p 22 without problem, and namd2 distributes fine.  Is there something else I need to
set for -pport (or -port, for that matter)?
> 

The failure is not from ssh, but rather from Xplor-NIH itself. Perhaps
dns isn't running on the compute nodes? 

> If I have a reasonable homology model, can I use it to start with?
> How do I do this? 
> 

You might use this structure to help generate initial likelihoods. John
can help more with this.

best regards--
Charles

Re: parallel marvin

On May 12, 2008, at 7:26 PM, David A. Horita wrote:

> If I have a reasonable homology model, can I use it to start with?   
> How do I do this?

You could, but I'd avoid it if at all possible.  If Marvin's able to  
converge without starting from a homology model, you're that much  
more certain of the structure in the end.

But if you're having trouble converging, you could try replacing the  
network analysis that's done during the initialMatch script with a  
structure-based filter, using your homology model.  I haven't had to  
do this myself in the cases I've tried, but I could put something  
together.  It shouldn't be a problem, but there are a few different  
ways to implement it, and I'm not sure which would be best.  I'll  
send out a separate announcement with the code later.

One last point, though, is that you could use your homology model to  
help you score how well the network filter and so forth are working.   
Take a look at the CVN example, and you'll see references to  
"referenceStructure" in the initMatch and summarizePass scripts.   
Replace the filename there with the filename of your homology model's  
PDB, and it'll give you some interesting statistics in the headers of  
your .peaks and .shiftAssignments files.

Hope this helps.

--JK

distance plot in pyXplor

_______________________________________________
Xplor-nih mailing list
Xplor-nih <at> nmr.cit.nih.gov
http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih