On 12/15/2015 09:24 AM, Johannes Loeffler wrote:
> I am looking at a range from 0.005 to 0.995, where as in between the two
> end values i have chosen a spacing of 0.05
> So the to perform mbar analysis I should perform my simulations rather
> with the cuda implementation?
>
> On 12/15/2015 04:17 PM, Hannes Loeffler wrote:
>> The problem with sander is that it can't sample the end-points but
>> rather only points up to 0.005 away (so the range is 0.005 to 0.995).
>> This necessarily means that there would be some missing energies and
>> thus some error if you only did MBAR analysis over that range. Sander
>> could principally estimate the MBAR end-point energies but proper
>> BAR/MBAR requires explicit sampling of all lambda points. The latter is
>> also a problem because the namelist variables bar_l_* only allow you to
>> set up an equally spaced array.
I'm probably quibbling on semantics here, but this is not entirely
correct. MBAR does not require explicit sampling at all lambdas, but
only that the energy of a snapshot at an unsampled lambda value can be
computed. I regularly use MBAR to extrapolate to endpoints when using
soft core potentials and only a very small/large lambda instead of 0/1.
This is not to say that the extrapolation is guaranteed to be /good/,
but it can be done. It is also often easier to just re-evaluate the
energies at other lambda values.
Regards,
Brian
>> What is the lambda range you are look at?
>>
>> Overall I don't know if there is a useful scheme to support this type
>> of analysis for sander. That's why I have decided to not support it.
>>
>>
>> On Tue, 15 Dec 2015 15:42:18 +0100
>> Johannes Loeffler <johannes.r.loeffler.gmail.com> wrote:
>>
>>> Thank you for your fast answer!
>>> I just seem to run into another problem;
>>>
>>> Note: BAR/MBAR results are not computed.
>>>
>>> is printed out whenever I try to run the script. I used the sander
>>> module of AMBER 14. Might that be a problem?
>>>
>>> Thanks again Johannes
>>>
>>> On 12/15/2015 02:46 PM, Hannes Loeffler wrote:
>>>> https://github.com/MobleyLab/alchemical-analysis
>>>>
>>>> or better yet my fork which has a few more bug fixes
>>>>
>>>> https://github.com/halx/alchemical-analysis
>>>>
>>>>
>>>> On Tue, 15 Dec 2015
>>>> 14:36:06 +0100 Johannes Loeffler <johannes.r.loeffler.gmail.com>
>>>> wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> I am trying to perform mbar analysis and was wondering if there are
>>>>> any tutorials on how to extract the output from amber out files. In
>>>>> the simulations ifmbar=1 flag was naturally included!
>>>>>
>>>>> Thanks for your input
>>>>>
>>>>> Joe
>>>>>
>>>>> _______________________________________________
>>>>> AMBER mailing list
>>>>> AMBER.ambermd.org
>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>> _______________________________________________
>>>> AMBER mailing list
>>>> AMBER.ambermd.org
>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
--
Brian Radak
Postdoctoral Scholar
Gordon Center for Integrative Science, W323A
Department of Biochemistry & Molecular Biology
University of Chicago
929 E. 57th St.
Chicago, IL 60637-1454
Tel: 773/834-2812
email: radak.uchicago.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Dec 15 2015 - 08:00:03 PST
