Firslty, there is credo http://marid.bioc.cam.ac.uk/credo a database set up in the Blundell lab, that deals with protein ligand interactions. That might be of interest.
Second, as more general points. I see your point Shane, I'd argue there is no great value in trying to distinguish artefacts from relevant ligands. Whether or not the crystallographer perceives something as biologically relevant any binding informs the community about the protein in some way. For his use case, might it be better to consider what the student wants to achieve with this work? For example using a minimum number of heavy atoms would remove a lot of buffer molecules and all metals. The advantage of this being you are now looking at all ligand like substances that a chemist might consider working with. If you then felt the need to get rid of say MES, then I think you would need to justify that. Is it because it's too high concentration? Then I'd ban any component above a certain concentration, if it's given to the PDB, but then I think you're missing valuable data. Or is it because it doesn't bind with any potency? Then I'd pick only ligands above a certain ligand efficiency, but that again would miss lots of data. It might be interesting however, to compare the interactions found from those two subsets, with that of the broader ligand world. Hope this helps! Best wishes, Anthony On 11 Jun 2015, at 20:55, Shane Caldwell <shane.caldwel...@gmail.com<mailto:shane.caldwel...@gmail.com>> wrote: Hmm, this makes me think of how upon deposition to the PDB, we describe the expected biological assembly that a crystal structure represents. This is a judgement call, but can be backed up with experimental data. A similar flag for physiological ligands as opposed to components of the crystallization solution (with just as much subjectivity) could be very useful for differentiating physiological versus artifactual ligands. Of course, much easier said than done. Shane Caldwell McGill University On Thu, Jun 11, 2015 at 3:14 PM, Lucas <lucasbleic...@gmail.com<mailto:lucasbleic...@gmail.com>> wrote: This week a bioinformatics PhD candidate was talking about his PDB analysis tool which was aimed at studying protein-ligand statistics and asked for some advice on what should he consider as "biologically relevant" or not. I told him many of the most common molecules he found were buffers, cryoprotectants, metals used for phasing, etc., and also said that some cases could be more complicated (e.g., PO4 is very common because of its use as a buffer, but in the same time it could be biologically relevant in many cases such as kinases or phosphatases). Well, I tried my best to list most "usually biologically irrelevant" I could remember, but now I wonder if that is something someone has probably thought about doing before and there's some article/database dealing with it somewhere. Any suggestions? Lucas