smbt

# Introduction # smbt (1,2,3) is a software for similarity searches of fingerprints. It takes a fingerprint database as an input, and builds an index consisting of multibit trees (MT) (2,3) as an output. The index of MTs are used for searching fingerprints similar to a given query. Standard pointer-based implementations of MTs consume a large amount of memory for large-scale fingerprints, which influences applicability of multibit trees. Recently, we presented several memory-efficient representations of MTs by using succinct data structures (1), and demonstrated their applicability to the PubChem database, one of the largest databases of chemical fingerprints. One might worry about a slow down of search time for reducing memory usage of MTs. We also demonstrated that the search time of our succinct representation of MTs was not so different from that of pointer-based multibit trees, which are presented in [our paper](https://sites.google.com/site/yasuotabei/wabi_2012_tabei.pdf?attredi