An improved heuristic algorithm for founder sequence reconstruction from SNP recombinants

The maximum fragment length problem (MFL) is an effective mathematical model for solving the founder sequence reconstruction problem. Roli et al. proposed a constructive heuristic algorithm (it is named as RHRC in this paper) for solving the MFL model and a tabu search method for further optimizing the solution obtained by RHRC. RHRC algorithm produces uncertain solutions by introducing stochastic information. In this paper, an improved algorithm I-RHRC is presented with the explicit aim of providing ascertained solutions. Instead of using stochastic values, I-RHRC takes advantage of some potential information, i.e. the proportion of 0 and 1 entries in the column of the founder matrix and that of the recombinant matrix, and some other heuristic information, to get ascertained values. Experimental results show not only that I-RHRC algorithm can get better solutions than RHRC algorithm, but also that the tabu search method based on I-RHRC outperforms that based on RHRC.