Skip to content

touchpoint_06 Dec 2023

Jump to bottom
Stephen Thomas edited this page Dec 6, 2023 · 2 revisions

Agenda

  1. Housekeeping and review TODO (15 mins)
  2. Discuss MD workflows for small molecule in solution state and solid/liquid interfaces (15 mins)
  3. Start defining user stories (30 mins)

Attendees

  1. Mike
  2. Stephen
  3. Steven
  4. Ankit
  5. Cameron

Meeting notes

  • We reviewed the User Story 1 that was created in the previous meeting and clarified why the homology model check was looking for >12 disulphide bonds

  • We increased general awareness on bespoke FF

    • Based on questions from Cameron, Mike created an issue on openFF about what happens when a higher level of theory is used for the same fragment
    • Cameron explained why different functional groups require different levels of theory , for e.g. a hydrocardbon chain does not require a high level theory compared to a dihedral next to an aromatic ring. Since some of the algorithms scale by computational compelexity O(N^7), it is very important to choose the level of theory wisely
    • We also talked about the semi-empirical method which is a simplification of the QM, but less accurate

  • We started talking about small molecule MD workflows

    • In development, it happens often when there are scenarios where it is very useful to have lattice energies handy for crystal structures. A fictional example is the project team finds the free form changes form to its hydrate form during a milling step. The team wants to know what the lattice free energy is at room temperature and a slightly higher temperature/pressure for the free form and the hydrate to see if this can shed some light on the form change. However, currently such values are not available unless someone explicitly calculates it. However, if we had an automated pipeline that routinely calculates descriptors such as the lattice free energy at a range of temperatures for every crystal structure that was added to the internal database and the results made available through a dashboard, such questions will become trivial to answer

Action Items

  1. Edit and beautify readme for protein fields - Mike
Clone this wiki locally