Mechanism of ConSurf
Slide by Eric Martz (2002 - 2009)
  1. Obtain amino acid sequence of 3D chain from PDB file specified by user.

  2. Run PSI-BLAST to find the most similar sequences in SWISS-PROT or UniProt.
    ConSurf uses up to 50 sequences (user can request more) with E < 0.001.
    ConSurfDB filters the sequences to reduce redundancy.

    Enolase (4ENL)

  3. Align sequences with MUSCLE (or CLUSTAL-W).
    Optionally, user may supply the sequence alignment.

  4. Construct phylogenetic tree from sequence alignment.
    Optionally, user may supply the tree.

  5. Calculate conservation scores (and confidence intervals) for each residue in the 3D chain using a Bayesian method.

  6. Conservation scores are normalized (to mean 0, SD 1) and binned into 9 grades.

  7. Color the 3D model by conservation grades in FirstGlance in Jmol or Protein Explorer. Gallery.
More Details