DM – Last year was very exciting – Automating fitting to the point where we can do whole experiments on the fitting process itself. Sage release, refitting LJ and valence terms without adding any regressions. This was a huge problem before, I spent a lot of my time fixing this sort of thing manually earlier in my career. Bespokefit coming online and promising results from early users. Getting around problems associated with ParmEd use by developing Interchange. Also promising developments on the horizon - Alternate functional forms to 12-6, where proof of principle work is one step from implementation/production.
JW – Industry benchmarking is also an impressive effort, where we have a package which gobbles up some molecules and returns performance results for Open Force Fields. Dotson did great work on coordinating and supporting this effort.
allows them to benchmark on molecules they actually care about/use, are proprietary
DM: opens up new conversations and conduits for working together to advance the FF, infrastructure; gets them more invested/engaged in the effort
JW – This also creates a way for pharma to run tests that they trust.
KCJ – really impressive, every time I need to write a report it’s hard because I have to leave things out! No shortage of exciting work that is being done, real advancements being made
2021 planned and actual infrastructure team time usage
JW – Project planning can be challenging for this kind of effort and I’m still learning how to do it. We’re less than 3 years old so I’m asking for some understanding in this department. I prepared a retrospective of my planning process for 2021 (attached spreadsheet). I tried to keep the number of main tasks per person low (up to 3). I spent more time on industry benchmarking than I initially anticipated, and I haven’t left enough time for organizational/management tasks. I’m trying to fix that for 2022.
Matt more or less was sticking to the planned schedule, focusing on Interchange.
Dotson is running QCArchive and submission management, and he’ll continue to do so. He also invested a lot of effort in industry benchmarks. He’ll spend more time on FAH in the coming year, enabling protein-ligand benchmarks.
Trevor Gokey is the OpenFF fellow working on force field/science support.
Ivan Pulido has worked with us during the summer before joining the Chodera lab; focused on biopolymer infrastructure development
Note that infrastructure team will need to spend more time on maintenance this year as core packages have increased in number and scope.
All “completed” items were verified and removed. Priority was assigned for about the first 1/3rd to 1/2 of new items. Roadmap revision will continue in the next all-hands meeting.