PSFgen with RNA-like nucleotide monophosphates

From: Aaron Larsen (alarsen_at_molbio.mgh.harvard.edu)
Date: Wed May 06 2015 - 11:01:26 CDT

Hi all,

I am attempting to parameterize some RNA like nucleosides. They are similar
to RNAs except with some groups swapped and some different bond orders. The
first step of this process is the generation of PSF files to go along with
the pdb files. I’ve run into several problems. To overcome these issues, I
decided to work with a well-established nucleotide, adenosine monophosphate
<http://en.wikipedia.org/wiki/Cyclic_adenosine_monophosphate> (AMP). If I
can get everything working for AMP, I should be able to proceed to other
similar molecules. This is my current workflow:

1) Isolate AMP from a crystal structure of an RNA duplex. I keep the
phosphate and oxygens on the 5’ end and then add a hydrogen the resulting
negatively charged phosphate oxygen and the 3’-OH. I do this as this is the
form that I estimate is best to parameterize later on.

2) I load the resulting PDB into VMD and open the PSFgen application.
I apply the CHARMM36 force field specific to nucleic acids and attempt to
guess chains. This results in 4 chains, as apparently those hydrogens and
the oxygen that I added are not handled by patches in the topology file. So
I modify the chain and attempt to generate a PSF.

3) The resulting PSF file looks very distorted. The wrong
connectivities are resulting in bonds between distal parts of the molecule.

If you could please suggest where I am going wrong or propose an
alternative strategy, I would really appreciate it! If I can get this
working well for AMP, my next goal is to generate initial topologies for
the modified nucleotide monophosphates in Paramchem, populate the topology
file with those guesses, and proceed from there.

Any advise you can give would be very much appreciated, thanks!

Best,

Aaron

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:21:07 CST