Re: [AMBER] [cpptraj] Nativecontacts failing with long trajectory

From: Juan Eiros Zamora <>
Date: Fri, 24 Jul 2015 11:21:30 +0100

Dear Dr. Roe,

> OK, based on your input I think I know what is happening here. You are
> using an averaged structure as a reference, so that structure is changing
> based on how many frames go into the average (which is why you see
> different contact pairs reported when you increase the # of frames).
I understand this, but is it normal that all of the contacts disappear
(as it happened for the fourth analysis)? Surely, if they are present
when I load in 500 ns of trajectory, they would also have to appear
(even at lower fractions) if I load 4 trajectories of 500 ns combined?
Anyway, I kind of managed to do the analysis with each independent run
separately and then averaging the results.

> Using an averaged structure as a reference is a bad idea in my opinion.
> They tend to be very unrealistic, particularly when there is significant
> internal motion (e.g. a rotating PHE ring will "flatten out" and become
> almost a straight line). You should use a real structure as your reference
> for 'nativecontacts', otherwise the contacts you come up with will likely
> not represent actual contacts. If your system has a crystal structure
> available that may be a good candidate, or perhaps a properly equilibrated
> structure.
I'm afraid I picked this up from a similar question that I asked here
some months ago ( The thing
is that what I really would like to do is record every distance between
all the atoms in the two masks, and if they fall below a certain
distance mark it as a contact. In the end for every distance pair I
would have a number ranging from 0 to 1 (0 not present in any frame and
1 present in all of them). So the closest thing that I found out in
cpptraj was the nativecontacts function. I thought that if I used as a
reference an average structure that sort of would have the behaviour
that I'm looking for. I'll change the script to compare it to an initial
equilibrated structure and see how much my maps change.

Many thanks for your help,


> Hope this helps,
> -Dan
> On Thu, Jul 23, 2015 at 1:59 AM, Juan Eiros Zamora <
>> wrote:
>> Hi,
>> I'm attaching the failed output files. The cpptraj (my version is V15.00)
>> commands I'm using are:
>>> parm top.prmtop
>>> trajin
>>> rms first
>>> average crdset average_structure
>>> createcrd loaded_trajs
>>> run
>>> crdaction loaded_trajs rms ref average_structure
>>> nativecontacts :1-89 :396-412 ref average_structure byresidue resout
>>> cmap_1.dat
>>> nativecontacts :1-89 :386-395 ref average_structure byresidue resout
>>> cmap_2.dat
>>> nativecontacts :1-89 :249-289 ref average_structure byresidue resout
>>> cmap_3.dat
>>> nativecontacts :249-289 :386-395 ref average_structure byresidue resout
>>> cmap_4.dat
>>> run
>> I've tried using an offset of 1000 when calling the trajin command, but
>> still getting the same errors. But on shorter trajectories the output files
>> are printed fine.
>> All the best,
>> Juan
>> On 22/07/15 21:53, Daniel Roe wrote:
>>> Hi,
>>> What version of cpptraj are tou using? Also, can you provide your cpptraj
>>> input and output? Thanks,
>>> -Dan
>>> On Wednesday, July 22, 2015, Juan Eiros Zamora <
>>>> wrote:
>>> Hi everyone,
>>>> I've got a large trajectory file obtained from different independent
>>>> runs and I want to use the nativecontacts command to analyze it.
>>>> I have 4 nativecontacts analysis to do, which work fine if I use a
>>>> subset of the data. But if I load the whole thing, the contact residue
>>>> pairs files come up incomplete (some have only a subset of the pairs,
>>>> and the last one only has the header). All those files are written
>>>> properly if I use a small set of the data, so I'm guessing it's not a
>>>> problem of the analysis per se.
>>>> How can I figure out the offset I have to use when loading the big
>>>> trajectory to minimise accuracy loss on the analysis? That's the only
>>>> alternative that comes to mind, but I would like to use all the frames
>>>> (even if it takes awhile) if that's possible.
>>>> Thanks in advance for any suggestion,
>>>> Juan
>>>> _______________________________________________
>>>> AMBER mailing list
>>>> <javascript:;>
>> _______________________________________________
>> AMBER mailing list

AMBER mailing list
Received on Fri Jul 24 2015 - 03:30:03 PDT
Custom Search