Key steps in dbclus

• From the first molecule in the list, to the last do:

  • First molecule becomes centroid ‘0’
    • Calculate T-similarity for each molecule to centroid_0
      • If T-similarity >= given tanimoto (T) – label molecule as cluster_0 member
        • Else get next molecule
    • Go back to top of the file and find first molecule that has not been ‘labelled’
      • Label that molecule as centroid_1
        • Go down the list and if molecule not labelled do as above
    • When finished
      • All molecules that have not been labelled mark as singletons

Notes: