I've done this before...if only I could remember....

In a previous (work) life I've done a lot of PLP coding. Then I moved to a different employer where I had to use K###E...

Now I am rediscovering PLP and try to do something I did several employers/years ago (and with no access to my old protocols I cannot remember how I did it......)

In order to make a selection of compounds from a large dataset, I cluster the molecules and now I want to loop over the clusters and pick a percentage of each of the clusters to get for example a 10% set which has the same relative clustered composition. Been looking at looping etc, but I am struggling to get it to work.

Does anyone have a protocol that does something similar? Or is there even one in the standard examples hidden under a non-obvious way ?

Thanks in advance