Excap: Maximization of Haplotypic Diversity of Linked Markers

• Published in: PloS one Vol. 8; no. 11; p. e79012
• Main Authors: Kahles, André, Sarqume, Fahad, Savolainen, Peter, Arvestad, Lars
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 07.11.2013; Public Library of Science (PLoS)
• Subjects: Algorithms; Bioinformatics; Coding Region; Computer science; Data reduction; Databases, Nucleic Acid; Deoxyribonucleic acid; DNA; DNA, Mitochondrial - genetics; Forensic engineering; Forensic Genetics - methods; Forensic science; Forensic sciences; Genetic Linkage; Genetic Markers; Genetics; Genome; Genomes; Haplotypes; Haplotypes - genetics; Humans; Identification; Laboratories; Markers; Mitochondrial DNA; Models, Genetic; Molecular biology; Mtdna; Multiculturalism & pluralism; Population; Sequence; New York; United States > US; Sweden
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0079012

Genetic markers, defined as variable regions of DNA, can be utilized for distinguishing individuals or populations. As long as markers are independent, it is easy to combine the information they provide. For nonrecombinant sequences like mtDNA, choosing the right set of markers for forensic applications can be difficult and requires careful consideration. In particular, one wants to maximize the utility of the markers. Until now, this has mainly been done by hand. We propose an algorithm that finds the most informative subset of a set of markers. The algorithm uses a depth first search combined with a branch-and-bound approach. Since the worst case complexity is exponential, we also propose some data-reduction techniques and a heuristic. We implemented the algorithm and applied it to two forensic caseworks using mitochondrial DNA, which resulted in marker sets with significantly improved haplotypic diversity compared to previous suggestions. Additionally, we evaluated the quality of the estimation with an artificial dataset of mtDNA. The heuristic is shown to provide extensive speedup at little cost in accuracy.