Fig. 1

Fingerprint processing in Sachem/Lucy. The left box shows a molecule from the compound database being broken into distinct substructural features by fingerprinting (only 4 are shown for demonstration); these are converted to keyword-like descriptors by hashing and stored in Apache Lucy. The right box shows the querying process. The query is broken into substructural features. These are then filtered to only include features with reasonable filtering power; the result is converted to keyword descriptors to build a text query, which is in turn run on Apache Lucy