pre-filter for querying very large collections of nucleotide sequences

 Raptor is a system for approximately searching many queries such as
 next-generation sequencing reads or transcripts in large collections of
 nucleotide sequences. Raptor uses winnowing minimizers to define a set of
 representative k-mers, an extension of the interleaved Bloom filters (IBFs) as
 a set membership data structure and probabilistic thresholding for minimizers.
 This approach allows compression and partitioning of the IBF to enable the
 effective use of secondary memory.