Atlassian uses cookies to improve your browsing experience, perform analytics and research, and conduct advertising. Accept all cookies to indicate that you agree to our use of cookies on your device. Atlassian cookies and tracking notice, (opens new window)
JR: how are charges assigned for the phosphate? Do =O and [O-] on the same phosphorous get the same charges?
JH: … can look into this to see if they’re symmetric.
MS: pretty sure charges are the same. Bond lengths would be from Sage 2.2.1.
JW: charges are the same.
JR: can check bond lengths.
CC: checked bond lengths, angles, torsions assigned to phosphates and they were ok for Sage.
LW – Are these the only TDs that this FF is trained on?
JH – Was only giving it the phosphate TD dataset that Patrick and Brent made. If we wanted others I’d need to rerun YAMMBS benchmarking. would be interesting to check how well it matches non-lipid phosphate geoms.
LW – Wondering how much nonbondeds are dominating the torsion profile vs. other FF components.
MS – If headgroups are getting too close and sticking, maaybe we could increase headgroup hetereoatom radius and keep them from sticking so much…. If anything lipid21 should have headgroups that are more attractive to ammonium… Charmm36 is more fluid than lipid21, and its phosphorous rmin_half was weirdly big.
MS: Lipid21 doesn’t use new LJ parameters.
JR – I thought NAGL was intended to fit AM1BCC? Why would this be different than AM1BCC?
JH – They’re supposed to match
LW – NAGL deviates more from AM1BCC reference as molecules get more highly charged.
JR – Are these lipids charged?
JH – The lipids in these slides are zwitterionic but net neutral.
JR – Proton transfers in quantum calcs?
LW – We’d need to check … connectivity change filter available in qcsubmit
CC:
Follow up on the phosphate LJ change: Tom Cheatham found that using the new LJ params for the non-bridging oxygens but keeping the old LJ params for bridging (i.e. ester) oxygens works better for nucleic acid backbones. This probably won't solve everything, but it might be worth trying if you have extra compute time.