I've just had a quick skim and there is a lot to check but is/can the public API preserved?

I've just had a quick skim and there is a lot to check but is/can the public API preserved?

Hi @johnmay ,
yes, all pre-existing public methods are still there and haven't been changed. The output might differ a bit, though, because of the saturation fix and other alterations to make the algorithm more "usable".

@rajarshi,
we saw that you originally implemented this algorithm way back when. Might I ask whether you had a (literature) reference for the algorithm? Or did you develop it yourself? Did you have a specific application in mind?

Hi @JonasSchaub as far as I can recall this was not based on any literature reference. I figured it was such a trivial method, that nobody would write it up! At this point I can't recall why I implemented it, beyond providing an alternative to the Bemis-Murcko fragmentation scheme

Thanks, @rajarshi, for the quick response!
We also couldn't find any literature about the algorithm itself, even though an approach like this is used in multiple places, e.g. MetFrag and some matched molecular pair algorithms.

@johnmay also did something similar to improve database searches (https://www.nextmovesoftware.com/talks/Mayfield_HoneyIShrunkTheDatabase_ICCS2025.pdf).

We are going to integrate it as an additional fragmentation scheme into our MORTAR application and then see what useful stuff we can do with it.

Sorry for delay, been busy. Looks good, thanks for the improvements.

@ToLeWeiss, I looked at the git history, and personally prefer a more linear history. It's hard to see what happens, with all those mergers. An alternative is to regularly use git rebase on your branch (/PR).

Thank you for taking the time to take a look at the changes @johnmay!

@egonw, thanks for the tip! I will be careful to rebase instead of merging in the future.