One first caveat I would say - "SAR by catalogue" can be very disappointing. In spite of the many millions of compounds available to buy, it is unlikely there will be a nice set of SAR a) based on the chemotype you have and b) exploring the SAR you would like to explore.
So for this reason - I would massively suggest you link up with a nearby chemistry lab if at all possible and carry out some simple enumerations (again if the chemistry is tractable) using custom chemistry.
However to actually answer your question. I would move away from methods like docking to search through a huge library like ZINC. They are computationally very expensive and you're not finding structurally related compounds using docking.
Instead you could use ligand-based methods to find compounds that present the same chemotype or pharmacophore to your ligands.
Chemotype - could be done using the ZINC website alone. Place your compounds in and then find all compounds that are say 90% similar in the zinc database, or for which your compound is a substructure.
Pharmacophore - many pharmacophore generation and screening packages exist. These can quickly search libraries of compounds for others that have the potential to bind to your site.
As a hybrid method between structure-based docking and ligand-based pharmacophore searching Schroedinger's e-pharmacophores would be a nice compromise
Hope this helps.
From: CCP4 bulletin board [[log in to unmask]] on behalf of Mo Wong [[log in to unmask]]
Sent: 16 June 2015 19:38
To: [log in to unmask]
Subject: [ccp4bb] Following up on fragment hits
I have started generating hits from a fragment library that has been screened by SPR, thermal shift and crystallography. We have a few potential allosteric binders that would, for selectivity reasons, be quite interesting if they show modulation of enzyme activity.
I am thinking about performing the follow up using a "SAR by catalog approach" - screening for higher affinity compounds by the above approaches and then validating these hits using to a bioassay.
My question: I'd like to mine the ZINC database for new compounds structurally related to my hits that I can screen. I am assuming this is usually done by a docking approach, but given I have co-structures I'm guessing there's a more guided approach I can use. Any suggestions (such as modules for Schrodinger or SYBL) that can take advantage of my crystal structures would be gratefully appreciated.