> If igb=1 is so bad,I'll choose other values,maybe IGB=2. What I plan to do is to add the implicit solvent effect during the simulation by turning on the IGB flag.
> Regarding the specified value of the IGB,I don't care so much as long as it can work.
there are a few really important things that you need to consider.
1) the specific igb flag makes a very big difference. unless you know
what the strengths and weaknesses of the different models are, you
might not want to use them at all. the differences between igb=1,2 or
5 can make the difference between success and years of wasted time.
2) running a successful simulation is not as simple as just getting it
to run without an error. i strongly suggest finding a peer-reviewed
article that reports a careful study that is similar to what you want
to do, and follow their methods. don't just turn on a flag and assume
that it will be ok.
finding relevant examples is really the only way to know which igb
flag might work best to solve the problems you want to solve-
otherwise you need to do a research project into whether these models
give accurate data for problems like yours.
3) for large systems, GB can be much, much slower than explicit
solvent. there are good reasons to sometimes use GB, but speed is not
one of them. are you sure that this is what you want to do? have you
compared the speed with that of using explicit solvent? people often
assume that GB is faster because there are few atoms- but this is
simply not always true.
the key idea is that all of our methods have strengths and weaknesses,
and none are best for every project. you really need to know which
method works best for YOUR type of project. I cannot stress this
enough- it's a main reason why the peer-reviewed literature exists (as
an archive of protocols and how well they work).
>
>
> 2010-05-06
>
>
>
> geyan
>
>
>
> 发件人： Carlos Simmerling
> 发送时间： 2010-05-06 21:47:46
> 收件人： AMBER Mailing List
> 抄送：
> 主题： Re: Re: [AMBER] can't make proper use of IGB flag, it cause"inputconversion error"
>
> I see- well in that case I still recommend against igb=1, it has
> serious and well documented problems.
> On Thu, May 6, 2010 at 9:39 AM, geyan <geyan.big.ac.cn> wrote:
>> I'm very sorry,I made a big mistake by saying implicit solvent to explicit solvent.What I meant to do is to simulate the interaction of my DNA and protein complex in spores,that clearly has no water.But with the helpful suggestions of others here,I decide to do the MD in implicit solvent effect and meet the problems I have issued.
>> Sorry to confuse you.Thanks for your help.
>>
>>
>> 2010-05-06
>>
>>
>>
>> geyan
>>
>>
>>
>> 发件人： Carlos Simmerling
>> 发送时间： 2010-05-06 19:18:45
>> 收件人： AMBER Mailing List
>> 抄送：
>> 主题： Re: [AMBER] can't make proper use of IGB flag, it cause "inputconversion error"
>>
>> I'm much more worried by what you are trying to do than the error messages
>> you are getting. combining GB with an explicitly solvated system is, in my
>> opinion, a bad idea unless you know exactly what you are doing.
>> I'm not sure what you mean by some of your statements.
>> why "out of necessity"? for a large system, GB is probably much slower per
>> step than using PBC and PME. by using GB and explicit water, you are getting
>> the inaccuracy and slowness of GB along with the viscoty of explicit water-
>> the worst features of both and probably no benefit. I emphatically recommend
>> against this.
>> when you say that you tested igb=1 and "it's OK", what tests did you run? we
>> have shown in published work that igb=1 combined with explicit water
>> performs very badly- the problem is that the solvation energy of the
>> explicit water in igb=1 is very far from correct. once again, I recommend
>> against this, and if I reviewed a manuscript that used this combination I
>> would probably recommend against publication.
>> so although you might be able to get these inputs to work, my advice is to
>> think about whether this really gives you a benefit, and do some careful
>> comparisons. this would not mean short MD showing that the system doesn't
>> fall apart, but something that relly tests the accuracy. also do timing
>> comparisons to make sure that this is actually faster than the methods
>> nearly everyone else uses (its hard to imagine that it would be, since you
>> should not be using a cutoff with GB and that will be very slow).
>> On Thu, May 6, 2010 at 4:30 AM, geyan <geyan.big.ac.cn> wrote:
>>> Hello all Amber users and Ross,
>>> Out of necessity,I have to do a MD under explicit solvent by turning on
>>> the IGB flag. I have tested the case of the Amber Basic tutorial with
>>> explicit solvent
>>> by setting IGB=1 and it's OK.In my system,a complex composed of DNA and
>>> proteins,I do MD by following the steps,first minimization,then heating
>>> stage,
>>> then density stage,then equilibration stage,at last,the production
>>> stage.But if I turn on the IGB,the MD will crash with the calculating
>>> extending.
>>> >From the error file produced,I got the following message:forrtl: severe
>>> (64): input conversion error, unit 9, file
>>> /home_soft/home/wujy/geyan/Amber/ABC_com_implicit/com_md5.rst
>>> Then I searched the archive of Amber,and found Ross has discussed with
>>> others about this issue,
>>> http://structbio.vanderbilt.edu/archives/amber-archive/2006/2736.php.Truely,mynewly produced .rst file has **** in it.But now the question is by simply
>>> following her instructions,that is changing the ntx=6 won't work because it
>>> will bring the error like this: forrtl: severe (39): error during read, unit
>>> 9, file
>>> /home_soft/home/wujy/geyan/Amber/ABC_com_implicit/density.rst.Regarding the
>>> NTXO,as far as I know,the default value in Amber10 is 1.
>>> So how to figure out this problem? I am very confused that if I don't
>>> turn on the IGB flag,and make other parameters the same as before,the whole
>>> process of MD will work well.
>>>
>>> The details of my input is in attach.
>>>
>>> Any suggestion is appreciated.
>>>
>>>
>>>
>>> --------------
>>> geyan
>>> 2010-05-06
>>>
>>> _______________________________________________
>>> 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 May 06 2010 - 07:30:07 PDT
