Just to add what Jiri wrote, force fields are not perfect and most
probably other parts need to be revised, too. One thing that needs to be
verified on a structure, first, is if its stable. A regular Watson-Crick
RNA will most probably stay stable (with the revised chi, alpha/gamma
corrections). If you go beyond this structure and start to simulated much
more structures, things will probably change.
We found out the problem with chi while trying to explain some
experimental results using computational methods such as Thermodynamic
Integration, which is a tool to calculate free energy changes of two
states. The system had a more complex structural feature than a regular
duplex; single-stranded RNA. It was not so easy to figure out which part
in a force field would cause problems in the single-strand simulations.
Experimental computational simulations were done in order to get a glimpse
on where the problem was lying. These dummy simulations actually
pinpointed me to revise the chi torsions. We then decided that we first
test on a more simple structure, RNA nucleosides. Later, GACC
single-strand RNA was benchmarked (against experimental results). Finally,
the original problem that included the TI calculations was tested. We also
revised the epsilon/zeta and beta torsional parameters in this paper
I. Yildirim, S. D. Kennedy, H. A. Stern, J. M. Hart, R. Kierzek, and D. H.
Turner, "Revision of AMBER torsional parameters for RNA improves free
energy predictions for tetramer duplexes with GC and iGiC base pairs," J.
Chem. Theory Comput. 8, 172 (2012).
There are two reasons why I wrote the above paragraph: You should use the
most correct force field in a simulation, even if the structural
features might be much more complex than a regular RNA duplex. A
riboswitch is definitely hard to simulate compared to a duplex DNA. It is
possible that the current force field is good enough to simulate the
riboswitch system, but I would not be surprised to see if it fails.
Figuring out where the problem lies will help the current force fields
to advance into a better scheme (in case the force field fails to predict
the properties of riboswitch).
The second reason is, even though a regular MD simulation will tell you
something about the system, there are different computational techniques
that you can use to verify/test/benchmark a force field. We used TI
calculations to get the free energy differences, but umbrella sampling MD
simulations combined with WHAM is another powerful tool to get free energy
surfaces that can be directly compared to experimental results. Currently
we are predicting properties of RNA loops using umbrella sampling MD
simulations, which is being compared to experimental results. Long story
short, computational methods in AMBER can be used to benchmark the force
fields in complex systems such as the case with riboswitch (if there are
experimental results that can be benchmarked).
Ilyas Yildirim, Ph.D.
-----------------------------------------------------------
= Department of Chemistry - 2145 Sheridan Road =
= Northwestern University - Evanston, IL 60208 =
= Ryan Hall #4035 (Nano Building) - Ph.: (847)467-4986 =
=
http://www.pas.rochester.edu/~yildirim/ =
-----------------------------------------------------------
On Wed, 18 Jul 2012, Asmita Gupta wrote:
> Dear Users,
>
> Recently, two new force field corrections for RNA chi torsion angles have
> been introduced i.e. OL3(Banas et al) and YIL (yildirim et al). While there
> are some papers which have tested these two parameters on individual
> nucleosides and -tides as well as short RNA structural motifs like
> kink-turns and tetraloops with time scale from 300-1000ns(approx), i just
> want to know whether these parameters have been tested on whole and intact
> structures like riboswitches or RNA aptamers. It has been proposed that
> both the parameters work well in the range of ~300-400ns long simulation,
> but if one tries to take a complete structure with all non-watson crick
> base-pairs and other unusual geometries into account, wouldn't the
> behaviour of the 2 parameters change?
>
> Also, if one wants to test these two parameters on a structure like
> riboswitch, then what should be the appropriate time scale for a simulation?
> _______________________________________________
> 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 Wed Jul 18 2012 - 09:00:04 PDT
