Computational and Statistical Analyses of Insertional Polymorphic Endogenous Retroviruses in a Non-Model Organism

Computation

Le Bao⁷  Daniel Elleder⁵  Raunaq Malhotra³  Michael DeGiorgio²  Theodora Maravegias²  Lindsay Horvath⁶  Laura Carrel⁴  Colin Gillin¹  Tomáš Hron⁵  Helena Fปryová⁵  David R. Hunter⁷  Mary Poss² 

[1] Department of Fish and Wildlife, 4034 Fairview Industrial Dr. S., Salem, OR 97302, USA; E-Mail:;Department of Biology, The Pennsylvania State University, University Park, PA 16802, USA; E-Mails:;Department of Computer Science and Engineering, The Pennsylvania State University, University Park, PA 16802, USA; E-Mail:;Department of Biochemistry and Molecular Biology, Penn State College of Medicine, Hershey, PA 17033, USA; E-Mail:;Institute of Molecular Genetics, Academy of Sciences of the Czech Republic, Videnska 1083, Prague, Czech Republic; E-Mails:;Department of Pathology, Johns Hopkins University, Baltimore, MD 21287, USA; E-Mail:;Department of Statistics, The Pennsylvania State University, University Park, PA 16802, USA; E-Mails:

关键词: endogenous retrovirus;    insertional polymorphism;    mixture models;    de novo clustering;    mule deer;    population history;   

DOI  :  10.3390/computation2040221

来源: mdpi

【 摘 要 】

Endogenous retroviruses (ERVs) are a class of transposable elements found in all vertebrate genomes that contribute substantially to genomic functional and structural diversity. A host species acquires an ERV when an exogenous retrovirus infects a germ cell of an individual and becomes part of the genome inherited by viable progeny. ERVs that colonized ancestral lineages are fixed in contemporary species. However, in some extant species, ERV colonization is ongoing, which results in variation in ERV frequency in the population. To study the consequences of ERV colonization of a host genome, methods are needed to assign each ERV to a location in a species’ genome and determine which individuals have acquired each ERV by descent. Because well annotated reference genomes are not widely available for all species, de novo clustering approaches provide an alternative to reference mapping that are insensitive to differences between query and reference and that are amenable to mobile element studies in both model and non-model organisms. However, there is substantial uncertainty in both identifying ERV genomic position and assigning each unique ERV integration site to individuals in a population. We present an analysis suitable for detecting ERV integration sites in species without the need for a reference genome. Our approach is based on improved de novo clustering methods and statistical models that take the uncertainty of assignment into account and yield a probability matrix of shared ERV integration sites among individuals. We demonstrate that polymorphic integrations of a recently identified endogenous retrovirus in deer reflect contemporary relationships among individuals and populations.

【 授权许可】

CC BY   
© 2014 by the authors; licensee MDPI, Basel, Switzerland.