FinIS: Improved <i>in silico</i> Finishing Using an Exact Quadratic Programming Formulation Algorithms in Bioinformatics
edited by: Ben Raphael, Jijun Tang
With the increased democratization of sequencing, the reliance of sequence assembly programs on heuristics is at odds with the need for black-box assembly solutions that can be used reliably by non-specialists. In this work, we present a formal definition for in silico assembly validation and finishing and explore the feasibility of an exact solution for this problem using quadratic programming (FinIS). Based on results for several real and simulated datasets, we demonstrate that FinIS validates the correctness of a larger fraction of the assembly than existing ad hoc tools. Using a test for unique optimal solutions, we show that FinIS can improve on both precision and recall values for the correctness of assembled sequences, when compared to competing programs. Source code and executables for FinIS are freely available at http://sourceforge.net/projects/finis/ .