GROMACS does not support TMD, and there is no mention of it in the Desmond manual, so I would suspect that it doesn't, either.

GROMACS does not support TMD, and there is no mention of it in the Desmond manual, so I would suspect that it doesn't, either.

Thanks, Justin and Stéphane for comments. Is it a good idea to go for targeted moleculardynamics when (i.e, you have the starting structure a crystallographically resolved PDB file and a homology modelled target protein coordinate file carrying exactally the same component amino acids and that target structure is based on certain observations and published data).

It's not a homology model if it's the same protein, only in a different conformation. In this case, you'd be proposing a putative pathway for conversion to a putative state. There are a lot of variables there. Maybe it will be informative, maybe not. That depends on the extent and quality of the experimental information used to construct the putative second state.

ok.. basically the target structure is a homology based structure as the experimental protein is known to transition from active to inactive state through this intermediate transition structure,, as I dont have the coordinates of this intermediate state with me so I have to model the structure through homology modelling using the coordinates of a homologous protein with whose structural pattern this transition state structure matches. I am doing this because I cant afford an unbaised MD of more than 15 micro seconds.

I understand the technique you're using, but I still wouldn't call the product, strictly speaking, "a homology model." It's merely an alternate conformation of a protein based on the assumption that it looks like another protein. Just my $0.02 in cautioning you on the clarity of such a statement.

The main point I was trying to make is that you are making compounded assumptions. Targeted MD can be a powerful tool in exposing transition pathways, but the pathway you discover depends heavily on the end states you define. You're making an assumption that one state looks a certain way, based on sequence (but not necessarily structural) similarity. If you have significant experimental corroboration that justifies that assumption, then perhaps you can make a strong argument. Not all reviewers might see it that way, and that's what I was trying to point out.

What might be worthwhile doing is to apply the same method to the homologous protein, if both of its end states are known, to uncover any common features of the transition pathways that may indicate any sources of error or bias.

Justin Lemkul.. Sir I have been reading your Bewan Lab turtorials on Gromacs and those are really good. Its pleasure getting advise from you. I have read your Article Practical considerations for building GROMOS-compatible small-molecule topologies.

Sir can we use manually the Drug topologies generated by Desmond OPLS 2005 for GROMACS OPLS 2001,,

And I have understood your comment thanks for correction

I don't really know anything about Desmond or the topologies it creates, sorry.

Thanks for your comments Justin and Stephane. I was wondering whether to use plumed for targeted MD in gromacs seeing as I have a Martini system generated by CHARMM-GUI and this has answered me that I am right in thinking that's the way forward.

Also, seeing your other discussion, I just thought I'd add my 2 cents. Even if the specific problem discussed isn't relevant anymore, there will be people like myself who come across this later.

My main point is that there are other methods if targeted MD doesn't work. There are many of them so I can't discuss them all here and will instead just mention a couple.

1) We have a method in Ivet Bahar's lab called Collective MD (CoMD) that explores the transition based on collective modes from elastic network model normal mode analysis, which is implemented using VMD, NAMD and our Python package ProDy. You need to install VMD, NAMD and ProDy separately and then add the comd directory to VMD but after that there is a GUI in VMD to handle setup and it produces a script that can be provided to VMD for running everything. This method actually uses targeted MD for small sections of the transition. You can find more information about it at http://prody.csb.pitt.edu/tutorials/comd_tutorial/

2) If you choose to use plumed then there is a method called metadynamics with path collective variables that you might want to try as well. I have never used it but I find it intriguing and probably will use it at some point. There is a tutorial at https://plumed.github.io/doc-v2.4/user-doc/html/belfast-2.html

The advantage of such methods is that you can explore the space of alternative paths and not just the initial path you define. Targeted MD can do that to some extent if you use it carefully but as Justin said it's maybe not the best thing to do if you're not sure about things.