Edinburgh Research Explorer

Using genomic selection to predict canine Leishmaniasis and identify susceptibility segments from genome-wide SNP data

Research output: Contribution to conferenceAbstract

Related Edinburgh Organisations

Original languageUndefined/Unknown
Pagesabstract P661
Publication statusPublished - 2011
EventPlant & Animal Genomes XIX Conference - San Diego, United States
Duration: 15 Jan 201119 Jan 2011

Conference

ConferencePlant & Animal Genomes XIX Conference
CountryUnited States
CitySan Diego
Period15/01/1119/01/11

Abstract

Leishmaniasis is a parasitic vector-borne disease affecting humans and animals. Human and canine leishmaniasis have a worldwide distribution and constitute a public health and veterinary problem. Epidemiological and candidate gene studies in humans, mice and dogs have established susceptibility to leishmaniasis depends on host genetic factors. We applied genomic selection to predict canine leishmaniasis development and to identify those genomic segments with significant influence on predictions. Samples were collected from Boxer dogs from a region where canine leishmaniasis is endemic. Based on medical records dogs were classified as either healthy, older than 4 years but with evidence of prior infection (n=87) or affected for the disease and diagnosed before 4 years (n=78). After data cleaning 100,909 SNPs were analyzed. A modified BayesB method was used with covariates describing lifestyle and genetic stratification. To measure the predictive power of the markers, permutation and cross-validation analyses were conducted. The correlation of the fitted values and the actual phenotype was (i) higher in the un-permuted dataset than in datasets for which phenotypes were permuted (0.99 and 0.69, respectively) and (ii) equal to 0.42 (p-value = 2.82 ž 10-8, CI=(0.28, 0.53)) in the cross-validation datasets. The area under the ROC curve was 0.73 and a fitted value threshold of 1.5 would produce the highest fraction of correct diagnosis predictions. In addition, three genomic segments on chromosomes 4, 5 and 20 had an increased SNP effect on the prediction of disease development. Further analysis with additional 56 Boxers and 43 German shepherd is in process.

Event

Plant & Animal Genomes XIX Conference

15/01/1119/01/11

San Diego, United States

Event: Conference

ID: 376990