Some CombiRIDGE Q&A questions.
Andrew Orry
CombiRIDGE is a GPU-accelerated platform for high-throughput 3D ligand docking method that combines anchor-based expansion with generative conformer enumeration and graph neural network scoring to efficiently explore vast chemical libraries.
Here are some recent Q&As about the method - you can also watch a webinar on CombiRIDGE here.
Q: Can I use a selected SDF as input for enumeration instead of the raw MEL.sdf from step #1?
A: Yes. You can use a selected SDF file as the input instead of the raw MEL.sdf output from step #1.
Q: How many anchors should I enumerate? Would 10 anchors with similar scaffolds be reasonable, or should I use a larger number such as 100 or 1000? What’s the upper limit?
A: The number of anchors to enumerate depends on how much chemical space you want to cover and how much time you want to spend. There’s no strict upper limit, but you certainly don’t want to enumerate billions of molecules. You can either:
-
Specify the number of anchors to enumerate (e.g. _combiRidge fr=1 to=10000 to enumerate the first 10,000 anchors — for Enamine this corresponds to roughly 10 million enumeration space), or
-
Use the mnmol parameter to limit the total space explicitly.
It’s also recommended to set a score cutoff (for example, -60 or -50 combined).
Q: What does the mnmol option refer to? Is it per anchor or total number of enumerated compounds? What if mnmol is smaller than the possible number of enumerations?
A: The mnmol parameter refers to the total number of enumerated compounds. Enumeration will stop once this total is reached, even if additional enumerations are possible per anchor.
Example:
_combiRidge mnmol=10000000 ...
This stops after 10 million total enumerations are reached.
Q: How should I set the mnhits option with respect to mnmol?
A: On top of the mnmol setting, you can provide mnhits=N, which defines how many top-scored compounds to keep in the final hit list. This lets you retain only the best-ranked subset after enumeration.