Is There A 'Best' Molecular Dynamics Software? How Is Each Different?

Question

[Hranjeev]

Molecular dynamics software predicts interaction between molecules. And, there are some molecular dynamics softwares that I have noticed such as Accelrys Discovery Studio and Tripos Sybyl (commercial), AutoDock, NAMD and etc.

[1] Firstly, what is the best all-in-one software?

I've seen many functions available in Discovery Studio and Sybyl. But, what does other free softwares has to offer - are they comparable? [2] Rather than aiding in workflow and all-in-one-ness what should be the main driving philosophy on picking what to learn in the field of in silico drug development. I understand that most softwares mostly overlap in providing established algorithms that do the analysis for us.

I am a novice bioinformatician and I have a biology/biotech background. [3] I want to learn one of them so which one would I choose to have among the tools at my disposal. Let's assume I have all the time at my disposal.

PS: So far I am planning to learn R, Linux, and maybe Perl as somewhere stated in another thread which seems to be the de facto for a person in bioinformatics. I already have some decent amount of Python and web framework.

Answer 1

[Khader Shameer]

Definition of "best" depends on various factors.

IMHO, best Molecular dynamics software is GROMACS - which is apparently missing from your list. I have used Sybyl and Accelrys Discovery Studion for several structure analysis pipe-lines, they are good for GUI based application - but not for a full MD simulation runs.

GROMACS is superior due to its strong user-base, parallel implementation, scalability etc. Most importantly, every other problem you encounter during an intensive MD run may have already been answered in one of their mailing list. I have not used NAMD - but heard it got similar features of GROMACS.

[Rm]

+1 for Gromacs software

Answer 2

[Mndoci]

There is no easy answer to your question. There is a set of codes out there which all have their strengths and weaknesses, ranging from force field, flexibility, licensing models, functionality and scalability. CHARMM, AMBER, NAMD and Gromacs are probably the four codes that you probably want to look into first, but there are others around that might meet your needs, e.g. Desmond. If you are an academic you can get them free or cheap. Commercial users might need commercial licensing, e.g. Discovery Studio (things could have changed since I managed that part of the product) uses the commercial version of CHARMM as it's MM/MD engine.

I used to be partial to CHARMM because it had some killer functionality and extensibility that the other codes just didn't have, but things have changed a lot in the past 5-10 years. If I had to pick one today, I'd go with NAMD (scale and force field) or Gromacs (fast, GPLed).

[Hranjeev]

Very insightful perpective. Thanks

Answer 3

[Suk211]

"Molecular dynamics software predicts interaction between molecules"

Not quite, it's the Force field which will determine the dynamics of the protein system you are interested in. You can use any package(GROMACS/CHARMM/AMBER/NAMD/LAMPPS) but at the end it will depend upon the force field and the water model( if you are not simulating in vacuum).

In fact all these packages lets you plugin your own force field to simulate your protein system as well. If you are looking for analysis part then definitely different packages will have their own tools to analyze the simulated trajectories. Our Groups prefer Gromacs because we can write our own tool in C. if you are Fortran Programmer then probably you can use AMBER and if you believe in Object oriented Programming then LAMMPS might be the most suitable package for you.

Answer 4

[Comcon1]

If "predicts interaction between molecules" is the center of you interest. It's not main question "what program to use". The main question - "how to calculate interaction". What do you want indeed: thermodynamics of interaction? kinetic of interaction? structure of interaction complex? Depends on your answer you can choose appropriate software and techniques.

Answer 5

[Delinquentme]

A really interesting MD option that I've been trying to get my hands on is the Anton MD cluster.

I'm lucky enough to have one located an hour away in Pittsburgh .. but still I need to get in touch with a researcher who has machine time on it.

Anton is pretty cool because its hardware designed specifically FOR MD, allowing it to simulate several microseconds

...where usual computation allows only for time spans in nanoseconds