Exact Multiple String Matching Problem for DNA Alphabet

Given a text T = t1t2 ... tn and a set of patterns P = {P1, P2, ..., Pr}, the exact multiple string matching problem (EMSMP) finds the ending positions of all sub-strings in T which is equal to Pi for 1  i  r. We regard all substrings in T and patterns in P as data points in an edit distance-based metric space. The data points in T are constructed into a vantage point tree (vp-tree) T. Then, EMSMP can be resolved by searching all points of P in T. We further enhance T into vpac-tree C (vp-tree with alliance cut capability), based on which more unnecessary branches might be cut off so that the searching efficiency could be improved. Experiments consisting of long texts and short patterns of DNA alphabet are conducted using the two proposed schemes and m-BNDM (the multiple pattern version of the well known BNDM approach). The computational results demonstrate the effectiveness and efficiency of our schemes.