Fig. 1

Workflow: (a) Partitioning: reference sequences are divided into the sets G¹ and G². Each reference is further partitioned into slightly overlapping windows w_i. (b) Database construction: the s smallest k-mers of each window are computed and inserted into the database. (c) Classification: a database is queried with the s smallest k-mers of a read. The returned hits are used to count the number of hits within each window. Target reference genomes are identified by high scores in the window count statistics. In case of several partitions, the top hits from querying each database need to be merged in order to assign a read to a reference genome. After all reads have been processed, coverage check and quantification are performed