2022_02_11 OpenMM/OpenFF check in

PE – I’m unable to see the meeting notes

• JW – We changed the structure of our confluence spaces, the new meeting notes must not be publicly visible. I’ll fix this after our meeting today.,

JW – OpenFF is switching to a vendored periodic table solution - Taking the symbols and atomic masses from OpenMM

• PE – Isotope masses would make more sense

• JW – True, but we’re trying not to rock the boat too much yet.

JW – Advice for how to “automagically” create residue info?

• https://docs.google.com/presentation/d/1sV4CQ5bORuAKy2q8TUJvaSVf2SI5Dy8HS8b-S7TsVis/edit#slide=id.g11153939175_0_2

• PE – Important to note:

  • At parameterization-time, OpenMM doesn’t try to make any judgements about relationships between chain IDs and mutually connected atoms. There is some notion of a difference between “internal and external bonds” (internal or external to a residue) 

  • At loading-time, chains are used to inform whether there should be peptide bonds between adjacent residues. At parameterization-time, chain info isn’t used. 

  • If noncontiguous atoms are part of the same chain, they’ll need to be part of different chains. 

  • If OMM writes out an OpenFF topology, standard residues won’t have CONECT records written. Check what happens if a nonstandard AA with a modified chemistry is exported. It will see the name “ALA” and omit writing the bonds, which will be a big problem if it’s not chemically alanine. 

  • Multiple OpenMM chains can share exactly the same name!

  • Two residues with the same number/resSeq value, in the same chain, are assumed to be AltLoc.