Re: [AMBER] QM/MM link atoms

From: Gustavo Seabra <gustavo.seabra.gmail.com>
Date: Thu, 2 Aug 2018 15:25:00 +0000 (UTC)

Hi,




Glad to know it's working now.




About the speed, all I can say is that probably >90% of the time is spent on the QM part, so if you need to speed it up, I'd check the QM software. Unfortunately, I have no experience with Orca to help there.




All the best,




--
Gustavo Seabra
De: emanuele falbo
Enviado: quinta-feira, 2 de agosto 11:59
Assunto: Re: [AMBER] QM/MM link atoms
Para: AMBER Mailing List
Thank Gustavo for replying. I moved on, basically the Li+ atoms were included in "file.mol2", and it didn't work. But now I let AMBER add Li+ around the molecule " addIons2 mol LI 27" , so that my the topology file had charge 24+ ( because the molecule has charge 3-). And , using adjust_q=2, the system is now neutralized and the calculation is running fine. However, I am using AMBER(serial) with ORCA( 24-CPUs) and it is really slow? Do you have any idea how to get it faster? I am treating the molecule at QM , and the Li at MM. Here is my "file.in" 300K constant temp QMMMMD &cntrl imin=0, ntb=1 cut=8.0, ntc=2, ntf=2, tempi=300.0, temp0=300.0, ntt=3, gamma_ln=1.0, nstlim=1000, dt=0.002, ntpr=1, ntwx=1,ifqnt=1 / &qmmm qmmask='.1-53', qmcharge=0, spin=2, qm_theory='EXTERN', ! Use external QM program qmshake=1, writepdb=1, adjust_q=2, qm_ewald=0, qm_pme=1 / &orc use_template=1 / Best regards, Emanuele 2018-07-31 13:12 GMT+01:00 Gustavo Seabra : > Hi, > > > > > AFAIK, Sander will only add link atoms in places where it needs to break a > bond between QM and MM regions. Also, this 'bond' must be defined in the > prmtop file, i.e., it must be a real bond in your system. > > > > > If you only added Li+ íons around the molecule as counterions, there > should not be a bond defined, and no link atoms added. > > > > > Also, notice that using Sander will *not* help with the charge problem. > The QM part is sent to the QM program 'as is' (apart from any link atoms), > and if it is highly charged, it will keep this charge in the QM program. > Remember that the Li+ íons will be passed only as point charges that do not > affect the molecule charge. > > > > > The message you are getting actually tells me that the definition of the > QM zone may not be getting what you intend, as apparently there remains no > MM atoms. Are the Li+ íons being passed as part of the QM zone? From what I > understand, they should *not* be included in the mask. > > > > > My suggestion is that you try to print the QM region and compare to what > you expect. There is a keyword, I believe it it something like > 'qmmm_writepdb' (or something like it, check the manual) that tells Sander > to write a pdb' file with the QM zone. It will include only the QM atoms > (plus any link atoms), but *not* the MM part. With that you can be certain > of what is going on. > > > > > HTH, > > > -- > > > Gustavo Seabra > > > > > > > > De: emanuele falbo > > > Enviado: terça-feira, 31 de julho 08:09 > > > Assunto: [AMBER] QM/MM link atoms > > > Para: AMBER Mailing List > > > > > > > Dear admins, I am running a QM/MM simulation interfacing AMBER to ORCA. My > system is composed by a molecule, treated at QM level and surrounded by Li+ > atoms, which I wish to treat at MM level. The problem is that SANDER keeps > saying me that adjust_q=2 requires natom-(nquant+nlink)>=1 . 1. I did not > specify to include link atoms, cause I don't know how to do it. " 300K > constant temp QMMMMD &cntrl imin=0, ntb=1 cut=8.0, ntc=2, ntf=2, > tempi=300.0, temp0=300.0, ntt=3, gamma_ln=1.0, nstlim=1000, dt=0.002, > ntpr=1, ntwx=1,ifqnt=1 / &qmmm qmmask='.1-53', qmcharge=0, spin=2, > qm_theory='EXTERN', ! Use external QM program qmshake=1, qm_ewald=0, > qm_pme=1 / &orc use_template=1 / " 2. This is must be because the charge in > file.prmtop is 24- but I used "qmcharge=0" ( because I want ORCA to run a a > calculation on a neutral system). 3. In fact, in creating the file.prmtop > for my molecule I used the Mulliken charges from an ORCA calculation, and > the total molecule charge is 3- . But, I want to surround it by 27 Li+ , > and for this reason the total charge will obviously be 24-. The thing is > that I can't get the Mulliken charge from ORCA for my molecule, because the > molecule 27- doesn't converge, reason why I am trying a QM/MM run to > distribute the Li around. Do you have any idea about how this problem can > be fixed ? Best Regards, Emanuele _______________________________________________ > 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 Aug 02 2018 - 08:30:04 PDT
Custom Search