set up environment
source /nfs/soft/dock/DOCK37/env.csh source /nfs/soft/www/apps/zinc15/envs/internal/env.csh
zinc-manage -e shell utils retrosynth analyze --smiles="Cc3ccn2nc(CCCC(=O)Nc1ccc(O)cc1C)cc2c3" --recursions=0 --show=successful
sample output and explanation
Line-1 (1 options) Successful Synthesis via negishi coupling in 1 steps Cc1ccn2nc(CCl)cc2c1 . Cc1cc(O)ccc1NC(=O)CCCl >> Cc1ccn2nc(CCCC(=O)Nc3ccc(O)cc3C)cc2c1 Reactant 1: Cc1ccn2nc(CCl)cc2c1 - Purchasable as ZINC000238540793 Reactant 2: Cc1cc(O)ccc1NC(=O)CCCl - Purchasable as ZINC000041725839
If you want a bit more diversity you could change --show to partial to get reactions where less than all (but more than none) of the reactants were available. This also may be useful if you want to use a different reaction. You can the either run similarity searches around the unresolved reactants or try to increase the --recursions to try a few more steps... That can get out of hand above 2 or 3 though (not that you'd want to do that).
step 2 - Gets ZINC IDS
- 1) ZINC000238540793
- 2) ZINC000041725839
I then ran a similarity search for building block compounds containing the reactive group for the left and right side of a Negishi coupling (no easy interface available for this yet)
These are left.smi and right.sim in /nfs/work/teague/MK/
These current results are at 70% Tanimoto because I misplaced the 50% Tanimoto downloads (doh!)
I then ran the reactor in DOCK37 ($DOCKBASE/ligand/reactor/react.py) with the RXN SMARTS for the Negishi coupling ( "[#6:1]-,:[#17,#35,#53].[#6:2]-[#17,#35,#53]>>[#6:1]-[#6:2]" ) and the left.smi and right.smi and filtered for duplicates. This creates synthetic.smi. Modify that to be better suited for db2 generation.
(all in the run.sh script in that directory).
Run database building using the DOCK37 pipeline.
For the new compound you showed me, I think we need to investigate the reactions and see which one is failing to validate.
I am building the db2 files now in that directory.