A Mixed Integer Linear Programming Algorithm for Plasmid Binning

The problem of analysing bacterial isolates in order to detect plasmids has been widely studied. With the development of Whole Genome Sequencing (WGS) technologies, several approaches have been proposed to bin contigs into putative plasmids. Reference-based approaches aim to bin contigs by mapping or comparing their sequences against databases of previously identified plasmids or plasmid genes. On the other hand, de novo approaches use contig features such as read coverage and length for plasmid binning. Hybrid approaches that combine both strategies have also been proposed recently. We present PlasBin a mixed integer linear programming based hybrid approach for plasmid binning. We evaluate the performance of several binning methods on a real data set of bacterial samples.