Closed tkosciol closed 4 years ago
I don't think that's feasible to assess given the time it might take to convert UniBoost
to UniBoost_hhsuite
(not available on http://gwdu111.gwdg.de/~compbiol/uniclust/2017_04/
)
@bctaylor maybe you'd be willing to look into this at some point? Reference data is here: https://uniclust.mmseqs.com and I'd recommend looking into the paper mentioned on the website.
HHsuite is installed on Barnacle in /projects/microprot/tools/hh-suite-3.0.0/build
It's not urgent, but it would be interesting to see if that's feasible and do we see any changes in performance.
obsolete with the release of the BigFatDB
Once the pipeline is ready, test the performance of HHblits MSAs with UniClust30 and UniBoost10/20/30 databases on contact predictions. Use a reference set of easy (PSICOV set) and hard (CASP11 set) targets to evaluate.