Development of Multiple Interval Mapping for Mapping QTL in Ordinal Traits
No Thumbnail Available
Files
Date
2005-01-20
Authors
Advisors
Journal Title
Series/Report No.
Journal ISSN
Volume Title
Publisher
Abstract
Though methods for characterizing quantitative trait loci (QTL) using continuous data have been well established, development of methods and programs for analyzing QTL in ordinal traits is still needed. In this study, we developed multiple interval mapping (MIM) for studying traits with binary/ordinal phenotypic values. We call the method bMIM. Based on the threshold model, this method assumes a continuous underlying liability for the ordinal traits. With traditional QTL models for continuous traits, the liability can be characterized. Incorporating multiple marker intervals and using maximum likelihood method, bMIM can fit multiple QTL simultaneously and obtain estimates for parameters such as QTL effects and positions. With further implementation, epistasis effects can also be detected and tested. By developing bMIM, we supply a new way to analyze QTL in ordinal traits and provide help in studying the genetic architecture of ordinal traits. In addition, we address several questions, including how various factors (such as number of categories) affect mapping results, whether it is suitable to apply QTL Cartographer/MIM to ordinal data directly (QTLB), and how well are the results from bMIM when compared with results from using continuous data (QTLC).
We have partially answered our questions using computer simulations. For example, higher heritability values and more categories increase the power and accuracy of the parameter estimation, and proportions of categories have little effects when categories are moderately divided. When the QTL number is low and the heritability is high, applying QTL Cartographer/MIM to ordinal data directly may yield results similar to those from QTLC and bMIM. Though containing less information than continuous data, ordinal data can still yield similar estimations to those obtained from continuous data, when the loss of information is not too high. We show that epistatic effects can be detected for simple cases.
We also applied our method to a real data set from a male hybrid sterility study. Using hybrids between Drosophila simulans and D. mauritiana, nineteen QTL were detected with five having positive additive effects and fourteen having negative additive effects. The dominance relationships between alleles from D. simulans (S alleles) and D. mauritiana (M alleles) were mixed, with eight M alleles being dominant and eleven being recessive. Interactions among QTL were also detected. The results, together with those from other methods, suggest that the factors affecting male fertility in backcross hybrids may be determined by multiple factors and complex QTL interactions may exist.
Description
Keywords
quantitative trait locus, QTL mapping, ordinal data, threshold model, maximum likelihood
Citation
Degree
PhD
Discipline
Genetics
