Fig. 2
From: Fractional hitting sets for efficient multiset sketching
supersampler’s sketching strategy. In order to build sketches, supersampler computes super-k-mers over the input sequence. Fingerprints are associated with each super-k-mer by hashing their minimizers to an integer, hence an integer per super-k-mer. Super-k-mers associated to sufficiently low integers are kept in the sketch. Super-k-mers are put into partitions according to their minimizer