Hi Thomas,
Thanks a lot for your information and advice! As you suggested, I tried to simulate an existing GNRA hairpin structure 5’-GCGAAAGC-3’ (
https://www.rcsb.org/structure/2K71). In a 1μs trajectory, the RMSD to the first frame is finally around 3Å and almost all hydrogen bond between G and C are stable during simulation, as shown below(every 100 steps 1 frame). 
Then I build a relaxed 5’-GCGAAAGC-3’ sequence in tleap and simulate it in the same workflow, the ssDNA quickly form a compact structure in 500ns simulation(end to end distance from 30Å to 5Å) but not normal GC hydrogen bonds. It seems an oxygen atom in phosphate group of 3’C forms hydrogen bonds with 5’G as shown below. Could it because of sampling issue?
I’m using the simulation setup from Oweida et. al ’s paper
https://pubs.acs.org/doi/full/10.1021/acs.jctc.0c00931 . This setup has good agreement with experiment for 30nt PolyT sequence.
I noticed in Shaw’s paper it mentioned Amber-bsc1 had a greater tendency to collapse for Long Single-Stranded DNA Molecules but they didn’t test Amber-bsc1 for DNA hairpins. Although their DES-Amber seems work for a similar 5’-GCGAAGC-3’ DNA hairpin...
Best regards,
Jiaan
> On Oct 30, 2022, at 9:41 PM, Thomas Cheatham <tec3.utah.edu> wrote:
>
>
>> I’m trying to simulate a self-complementary oligonucleotide (5 -CTTATGGGCCCATAAG-3 ) by bsc1+TIP3P in 100nM NaCl,
>> for 500ns. However, I didn’t observe any secondary structures in the trajectory. I’m wondering if there's anything
>> wrong in my production run settings and how do I fix it. The md.in file is attached.
>
> Whether a DNA single strand will fold into a stem loop structure will
> depend on the force fields and also the sampling. Before doing any
> simulation work on this system, I would peruse the literature to see what
> others have done to date.
>
> Earlier work by Chen and Garcia used replica exchange to watch the RNA
> UUCG tetraloop to fold w/ 64 replicas each for 300-400 ns.
>
> Work by Chakraborty and co-workers in multiple papers looked at relatively
> short 100 ns simulations of different sequences with the CHARMM force
> field (c27) and saw collapse but not really stem-loop formation except in
> a few cases; really more disordered collapse.
>
> DE Shaw - new force field for single strands -
> https://pubs.acs.org/doi/10.1021/acs.jpcb.1c10971
>
> As pointed out in that Shaw paper, the "AMBER" and CHARMM force fields
> tend to generate highly compact single stranded structures; however we
> have also seen in shorter single strands metastabilty of the B-DNA like
> structure on the 500 ns time scale. See also results of Pollack and
> Onufriev and also quadruplex folding work by Sponer and colleagues.
>
> I personally would build the sequence into a stem loop structure (based on
> a couple of the tetraloop structures available) and see if it is stable in
> MD simulation or not, however as I do not think there is a structure known
> for the GGCC tetraloop, I would probably study a known tetraloop structure
> like GNRA (not UUCG which appears to have troubles with the AMBER force
> fields), first.
>
> --tec3
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Nov 12 2022 - 14:30:03 PST
