Fast Construction of Suffix Arrays for DNA Strings

To perform fast searching in massive data such as DNA strings, the most efficient method is to construct full-text index data structures of given strings. The widely used full-text index structures are suffix trees and suffix arrays. Since the suffix may uses less space than the suffix tree, the suffix array is proper for DNA strings. Previously developed construction algorithms of suffix arrays are not suitable for DNA strings since those are designed for integer alphabets. We propose a fast algorithm to construct suffix arrays on DNA strings whose alphabet sizes are fixed by 4. We reduce the construction time by improving encoding and merging steps on Kim et al.[1]'s algorithm. Experimental results show that our algorithm constructs suffix arrays on DNA strings 1.3-1.6 times faster than Kim et al.'s algorithm, and also for other algorithms in most cases.