You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We recently came across a paper that cited the toolkit as
AM1/BCC partial charges were assigned using the OpenFF toolkit with default
settings.
Unfortunately there are two (possibly three, soon) default settings. RDKit/AmberTools and OpenEye generate slightly different charges for many molecules (#447 and many other discussions) based on what is installed and licensed, and soon NAGL will introduce a third backend. Having code associated with paper(s) helps but doesn't necessarily solve this issue if it's not clear what packages are installed at runtime.
We can't enforce citation guidelines, but I suggest we should nudge people towards including critical information like the backend used for charge generation. Are there other details that are this important (read: text that is necessary to reproduce results, even when code is included)?
The text was updated successfully, but these errors were encountered:
We recently came across a paper that cited the toolkit as
Unfortunately there are two (possibly three, soon) default settings. RDKit/AmberTools and OpenEye generate slightly different charges for many molecules (#447 and many other discussions) based on what is installed and licensed, and soon NAGL will introduce a third backend. Having code associated with paper(s) helps but doesn't necessarily solve this issue if it's not clear what packages are installed at runtime.
We can't enforce citation guidelines, but I suggest we should nudge people towards including critical information like the backend used for charge generation. Are there other details that are this important (read: text that is necessary to reproduce results, even when code is included)?
The text was updated successfully, but these errors were encountered: