Figure 3:

Application of the BWT algorithm to the string ‘AGGCT$’, where ‘$’ marks the end of the string and is lexicographically smaller than all other characters. First, a suffix array is created; afterward, it is sorted lexicographically. After the BWT has been performed, only the last column of characters and the order of the starting positions (each has the same length as the string) within the original string are saved. This reduces the memory requirements to a linear scale relative the length of the string.