On 2 Dec 2019, at 11:32, Greg Landrum 
<greg.land...@gmail.com<mailto:greg.land...@gmail.com>> wrote:

HI Peter,

On Mon, Dec 2, 2019 at 10:11 AM Peter Schmidtke 
<peter.schmid...@discngine.com<mailto:peter.schmid...@discngine.com>> wrote:

I wanted to know whether there was an easy way in rdkit to decompose a product 
to its reactants using an  rdkit smarts-based reaction?
Basically I’m looking for the runReactants function but the other way around.
Is there any chance this would work with a reverted reaction or is this just 
wishful thinking?

Nice question. And, as with any good question, the answer is: "It depends".

It's pretty straightforward to reverse a reaction by creating a new one and 
copying the reactants/products from your example in as products/reactants in 
the new reaction. This will work as long as you don't use query features in the 
specification of the reaction.


Yep I saw I could easily redefine a reversed reaction. What I couldn’t find was 
that this also sort of works for reaction smarts.

Unfortunately it stops working sensibly pretty quickly once you introduce query 
features.

Here's a gist showing some of this.
https://gist.github.com/greglandrum/5ca4eebbe78f4d6d9b8cb03f401ad9cd

I'm having a hard time coming up with even a hack solution that isn't a ton of 
work, but I'll think about it a bit more. There may be a way to make this work 
on the C++ side.

To help with this, can you give a specific example of the kind of reaction 
you'd like to reverse?



We’re currently working on an algorithm to make that work in a purely pythonic 
way (for now). Let’s see if we can come up with something sensible and good to 
have your simple example. It is exactly the type of use case we have in mind.
For instance, take these reactions here (maybe they’d need a bit of adaptation 
& refinement for rdkit) : https://pubs.acs.org/doi/abs/10.1021/ci200379p (SI : 
https://pubs.acs.org/doi/suppl/10.1021/ci200379p/suppl_file/ci200379p_si_002.pdf)
 .

thanks a lot for the gist, didn’t know it was possible this way already :)

Peter



_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Reply via email to