Add ARC-MOFs?

[kjappelbaum]

https://chemrxiv.org/engage/api-gateway/chemrxiv/assets/orp/resource/item/62e04636cf661270d7b615c1/original/arc-mof-a-diverse-database-of-metal-organic-frameworks-with-dft-derived-partial-atomic-charges-and-descriptors-for-machine-learning.pdf

[kjappelbaum]

I like

• they have many structures
• they have some labels - also for some processes
• there are some sanity checks

I do not know

• how much sense DFT-derived charges on unoptimized structures make
• how much use the diversity analysis bring us - it always depends on the application
• what to do with one more screening with different settings -> we need to work on ML methods to more efficiently combine data from different sources
• I couldn't figure out (i didn't read carefully) what their process model is (you have a clue @FMcil - they just seem to say things like "VPSA" but do not cite anything and do not say how the process looks like?)

[Andrew-S-Rosen]

I can answer this:

• how much sense DFT-derived charges on unoptimized structures make

In Figure S5 of our QMOF paper, we investigated the change in DDEC6 charge before and after relaxation. The plot is below.

To my surprise, it does not make a big difference. About 90% of the partial charges are within +/- 0.05 units before and after relaxation. I don't know if this extends to other charge partitioning schemes, but it very well might.

However, I should note, the unoptimized structure here is from experiments -- not from a UFF relaxation or similar.

[kjappelbaum]

Thanks for bringing this up! I thought that I had seen this kind of plot before but I didn't remember where - of course, it was yours!

[kjappelbaum]

@FMcil do you have a clue what their labels (https://zenodo.org/record/6908728) mean? I'm in particular wondering about ssp and afm.