Skip to main content
Fig. 4 | Journal of Cheminformatics

Fig. 4

From: Chemoinformatics and structural bioinformatics in OCaml

Fig. 4

Using a bisector-tree (BST) to accelerate similarity searches on a database of \(10^6\) PUBCHEM molecules. Molecular encoding is Faulon’s signature molecular descriptor [80] (an unfolded-counted fingerprint) with height equal to one bond and parametrized over MOL2 atom types. The database is searched for all molecules with Tanimoto to query \(\ge 0.99\) (left; t = 0.01) or Tanimoto \(\ge 0.8\) (middle; t = 0.2). The brute force version is shown on the right. 50 molecules from the database were selected randomly to serve as queries. Creating the BST (database indexing) took approximately 5 min using a single core of our desktop computer

Back to article page