Inferring bond breakage in a reaction from InChI strings of reactans?

Question

I'm currently confronting a problem:

I would need to enumerate all chemical bond breakages that occur during a given chemical reaction, including the type of bond and the atoms participating in the bond. This is straightforward to do by visual inspection if one has, say, the 2-D structures of the reactants and products of a given reaction.

My problem is that I have a list of several thousand reactions, and I need a computer to do the job for me. To this end, I got the InChI (and SMILES) representations of all molecules participating in each reaction. At first, I thought that it would be rather easy to write a program that identifies and enumerates the broken bonds in the reactants (given the InChI strings), but by now I'm starting to become desperate.

Perhaps someone knows any software able to do this?

Answer 1

There is the Reaction Decoder Tool (RDT on github) that can - from the README - do:

2. Reaction Annotator (Extract Bond Changes, Identify & Mark Reaction Centres)

EDIT: It potentially could have InChI support, but a quick glance at this file suggests that it currently doesn't...

In fact, that would require the CDK/RDT to implement the rInChI format

The input formats includes reaction smiles which should be possible to generate from your smiles strings. Looks like you just jam a ">" inbetween reactants and products with a "." separating the molecules in each set.

which sounds like what you want, I think. Also it should be able to handle those volumes as it has been developed to work with reaction databases at the EBI.

(Full disclosure : I worked on the project for a little while :) but it does have InChI support - via the CDK).