Use custom pdb database and integration of AlphaFold db ?

[justine...@gmail.com]

Dear Harry,

I work on very divergent protein without entries in the PDB and I was wondering if I could use Phyre2 on custom pdb structures (obtained from modelling or ab initio predictions) ? 

Moreover, I was wondering if you would consider adding the AlphaFold db to the PDB to enrich the number of structures available for comparison using Phyre2...

Thanks for your help and thanks for your fantastic tool which is so helpful and easy-to-use !

Justine

[Powell, Harold]

Hi Justine

Sorry about the delayed reply!

Q1 - yes, you can use the one-to-one threading method available in "Expert" mode to use any PDB that you have access to, however it was obtained. AlphaFold models (as PDB files, we don't use mmCIF files yet) work fine with Phyre2.

Q2 - not as such, but probably in part. In order to keep the turnaround time reasonable, we use a truncated version of the PDB in our "fold library" - at the moment it's about 70,000 chains (including a few domains from when we used SCOP directly) compared to ~180,000 entries in the PDB (since many have multiple chains, we would have to consider rather more entries). The AlphaFold DB is currently around 365,000 entries (and is "soon" to increase to ~130,000,000) so it would make our search take a rather long time.

I'm looking at what we can do with the AlphaFold DB to make it (or parts of it) usefully accessible to Phyre2 users, but this is not a small job!