A Partition-Based Suffix Tree Construction and Its Applications

Abstract

A suffix tree (also called suffix trie, PAT tree or, position tree) is a powerful data structure that presents the suffixes of a given string in a way that allows a fast implementation of important string operations. The idea behind suffix trees is to assign to each symbol of a string an index corresponding to its position in the string. The first symbol in the string will have the index 1, the last symbol in the string will have the index n, where n = number of symbols in the string. These indexes instead of actual objects are used for the suffix tree construction. Suffix trees provide efficient access to all substrings of a string. They are used in string processing (such as string search, the longest repeated substring, the longest common substring, the longest palindrome, etc), text processing (such as editing, free-text search, etc), data compression, data clustering in search machines, etc. Suffix trees are important and popular data structures for processing long DNA sequences. Suffix trees are often used for efficient solving a variety computational biology and/or bioinformatics problems (such as searching for patterns in DNA or protein sequences, exact and approximate sequence matching, repeat finding, anchor finding in genome alignment, etc). A suffix tree displays the internal structure of a string in a deeper way. It can be constructed and represented in time and space proportional to the length of a sequence. A suffix tree requires affordable amount of memory. It can be fitted completely in the main memory of the present desktop computers. The linear construction time and space and the short search time are good features of suffix trees. They increase the importance of suffix trees. A suffix tree construction process is space demanding and may be a fatal in the case of a suffix tree to handle a huge number of long DNA sequences. Increasing the number of sequences to be handled, due to random access, causes degrades of the suffix tree construction process performance that uses suffix links. Thus, some approaches completely abandon the use of suffix link and give up the theoretically superior linear construction time for a quadratic time algorithm with better locality of reference.