Number of reads of fixed lengths. Panel (a) depicts the redundant dataset while panel (b) shows the length distribution of unique sequences. 20, 21 and 24-mers are the most abundant sequences, while only a small number of reads are longer than 24 bases. 24 and 21-mers are the most frequent sequences in the non-redundant dataset. Note that those sequences with less than 18 bases have been removed from the dataset.