Re: [AMBER] DIfferences between thermostats

From: Gustavo Seabra <gustavo.seabra.gmail.com>
Date: Thu, 24 Nov 2016 10:23:58 -0300

OK. That assumes that there is “some” force that will drive this motion, even if it is not imposed by you. What I *guess* is that the thermostat will add random pushes, that happen in random directions. When averaged out, my guess is that this effect will cancel itself out, and the only motion you see would be due to this external force.

The idea of “net flux” is something like:

1. Imagine your system aligned so that the expected flow is in the ‘z’ direction.
2. Measure the number of water molecules that cross the ‘xy’ plane in the ‘+z’ and ‘-z’ ways. (Call them N+ and N-, for example)
3. The net flux would be defined by: flux = (N+) - (N-)

So, my suspicion is that, assuming the thermostat effect is random, its contribution to N+ should be the same as to N-, thus cancelling out when you calculate the flux.

Of course, that all depends on the initial assumption that the thermostat effect is truly random.


Gustavo Seabra.


> Em 24 de nov de 2016, à(s) 10:10, Karolina Mitusińska (Markowska) <markowska.kar.gmail.com> escreveu:
>
> Hi Gustavo,
>
> so in that case - yes, I expect that water will travel through the protein.
>
> Best regards
> Karolina
>
> 2016-11-24 13:58 GMT+01:00 Gustavo Seabra <gustavo.seabra.gmail.com>:
>
>> Hi,
>>
>> What I meant is, do you expect the water to naturally move from one side
>> of the protein to the other, like in a water pump? (That’s what I
>> understood from “flow”).
>>
>> —
>> Gustavo Seabra.
>>
>>
>>
>>> Em 23 de nov de 2016, à(s) 19:45, Karolina Mitusińska (Markowska) <
>> markowska.kar.gmail.com> escreveu:
>>>
>>> Oh, OK, so I misunderstood Gustavo's question.
>>> In that case: no, I'm not adding any external force, I want to observe
>>> water moving from one place to another (I thought this is related with
>>> directionality) and interacting with some amino acids.
>>> So, please correct me if I'm wrong, there is no such thing like "the best
>>> thermostat" to make water molecules move like "in real life"?
>>>
>>> Best regards
>>> Karolina
>>>
>>> 2016-11-23 23:43 GMT+01:00 Karolina Mitusińska <mitusinska.gmail.com>:
>>>
>>>> Oh, OK, so I misunderstood Gustavo's question.
>>>> In that case: no, I'm not adding any external force, I want to observe
>>>> water moving from one place to another (I thought this is related with
>>>> directionality) and interacting with some amino acids.
>>>> So, please correct me if I'm wrong, there is no such thing like "the
>> best
>>>> thermostat" to make water molecules move like "in real life"?
>>>>
>>>> Best regards
>>>> Karolina
>>>>
>>>> 2016-11-23 23:04 GMT+01:00 David Cerutti <dscerutti.gmail.com>:
>>>>
>>>>> I think what Gustavo is getting as is a question of whether there is
>> some
>>>>> external force that is driving the water molecules through some
>> channel in
>>>>> the protein. If you're just looking at water coming and going as it
>>>>> pleases, that's an equilibrium process and Gustavo's guess is a good
>> one:
>>>>> the thermostats will add randomness that could help your sampling rate,
>>>>> but
>>>>> with a detailed energy surface like the one present inside a protein
>>>>> cavity
>>>>> or channel it's anyone's guess as to how a particular thermostat will
>>>>> affect the diffusion constant of water. What I can tell you is that if
>>>>> you
>>>>> run ntt=3 with gamma_ln=3.0 (Langevin thermostat, three frictionless
>>>>> collisions per picosecond), you should see the diffusion constant of
>> pure
>>>>> SPC/E water go from 2.4 cm2/s to about 1.7 cm2/s. The first is
>> correct,
>>>>> the second is not, but of course the density, heat capacity, and most
>>>>> other
>>>>> properties of the water remain unaffected. Again, it's veyr hard to
>> say
>>>>> what a thermostat like that would do to water inside a protein
>> channel. I
>>>>> couldn't say whether our water models are generally reliable when it
>> comes
>>>>> to residence times of water on the surface or inside of proteins.
>>>>>
>>>>> On Wed, Nov 23, 2016 at 4:26 PM, Karolina Mitusińska (Markowska) <
>>>>> markowska.kar.gmail.com> wrote:
>>>>>
>>>>>> Yes, there is directionality involved in these analysis.
>>>>>>
>>>>>> 2016-11-23 22:22 GMT+01:00 Gustavo Seabra <gustavo.seabra.gmail.com>:
>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>>> Em 23 de nov de 2016, à(s) 15:54, Karolina Mitusińska (Markowska) <
>>>>>>> markowska.kar.gmail.com> escreveu:
>>>>>>>>
>>>>>>>> 2016-11-23 17:45 GMT+01:00 Gustavo Seabra <gustavo.seabra.gmail.com
>>>>>> :
>>>>>>>>
>>>>>>>>>
>>>>>>>>>> Em 23 de nov de 2016, à(s) 05:14, Karolina Mitusińska (Markowska)
>>>>> <
>>>>>>>>> markowska.kar.gmail.com> escreveu:
>>>>>>>>>>
>>>>>>>>>> And what about the hydrophobic properties of amino acids? Does the
>>>>>>>>>> thermostat affect these properties?
>>>>>>>>>
>>>>>>>>> It should not. The hydrophobic properties are a function of the
>>>>> force
>>>>>>>>> field you use.
>>>>>>>>>
>>>>>>>>> Now, thinking about it again, you mentioned you are interested in
>>>>> the
>>>>>>>>> “flow” of water through the protein. In this case, depending on how
>>>>>> you
>>>>>>>>> define “flow”, it is likely that the choice of thermostat will make
>>>>>>> little
>>>>>>>>> difference. Yes, the thermostats affect the dynamics of water
>>>>>> molecules,
>>>>>>>>> but it in all directions, which could lead to a fortuitous
>>>>>> cancellation
>>>>>>> of
>>>>>>>>> errors when you calculate the “net flow”.
>>>>>>>>>
>>>>>>>>
>>>>>>>> Could you tell something more about how the thermostat could affect
>>>>> MD
>>>>>>>> results? And the "net flow”?
>>>>>>>
>>>>>>> Others could give more details here. But, basically, the thermostats
>>>>> give
>>>>>>> random “pushes” to the molecules, adjusting the velocities so as to
>>>>>> conform
>>>>>>> to a distribution valid at some given temperature.
>>>>>>>
>>>>>>>> By "water flow" I mean something like - watching how water molecules
>>>>>>> travel
>>>>>>>> through the protein and which amino acids are involved into this
>>>>>> process.
>>>>>>>
>>>>>>> Is there a directionality involved? Or do you mean just the residence
>>>>>> time
>>>>>>> of the water molecules around some residues? By “flow” I assumed some
>>>>>>> preferencial direction.
>>>>>>>
>>>>>>>> So could the thermostat change something here?
>>>>>>>
>>>>>>> My point is that the thermostats will affect how the water molecules
>>>>>> move,
>>>>>>> but the effect will be random and in all directions. So, depending on
>>>>> how
>>>>>>> you measure the flow, in the average you could have some error
>>>>>> cancellation
>>>>>>> and not notice any difference between the different thermostats.
>>>>>>>
>>>>>>> Gustavo.
>>>>>>> _______________________________________________
>>>>>>> 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
>>
> _______________________________________________
> 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
Received on Thu Nov 24 2016 - 05:30:02 PST
Custom Search