Is it possible to develop a node which will take an input structure and search against a second input table of structures and identify the structure(s) which have the largest MCS to the input structure. It would be particularly beneficial if you could choose how many of the structures with the largest MCS to return.
This can be useful as a measure of identifying nearest structures to a hit molecule, or molecule of interest, or even using a key structural motif for activity and want to find other molecules that closely hold that motif.
What are your thoughts?
That's a nice idea. I was thinking about an MCS node for quite a while but could never make up my mind which output it should create. I think I could setup a MCS node that returns the MCS between a reference and the current entry for a data table and also the size (number of atoms) for the MCS. Thus you could sort by MCS size, and handle cases where you have different MCS of same size and so on. I'll write such a node in the next hours and send you a download link.
Because this node will require MOE the request is still open if somebody from the free software community would like to give it a go ;-)