Notes from 01 24 2020 Skype Call

Present: Stephanie, Henry, Blake, Gydo, Daniel

• Main question: sampling or scoring

  • To examine sampling: look at qfit residue output folder. Additionally, flags to write out additional conformations (lessen scoring metric)
  • must sample bond samples (Henry)

• qFit3 now uses composite omit maps rather than 2Fo-Fc (these are NOT re-refined automatically)

• qFit3: now no clash detection

• qFit3 should ignore density from neighboring residues

  • how is density scaled? (can output intermediate maps)
  • could we require that all atoms are 'good enough' density? W/local density measures
    • see Gydo's papers

• Differences between qFit1 and qFit3

  • used to have separate residue classes
  • short sidechain vs. long sidechain build-up -> which of these still exists?

• qFit3 now tries many more DoF at once then bumps down to 1DOF

  • this can also cause some memory problems

• composite omit map now used & soft clash (remove density)

  • lose aniso B
  • now random around CB (still in code but not implement at the moment)?

• Goal of scaling:

  • p_cal - p_obs
  • could re-visit how this is done

• MIPQ

  • BIC is now on score versus previously was on cross correlation
  • qFit1 used discrete cut offs based on resolution, but qFit3 now uses BIC

• Meet in the summer?