[Frontiers in Bioscience E4, 2670-2682, June 1, 2012]

A score-statistic approach for determining threshold values in QTL mapping

Chen-Hung Kao1, Hsiang-An Ho1

1Institute of Statistical Science, Academia Sinica, Taipei 11529, Taiwan, R.O.C.

TABLE OF CONTENTS

1. Abstract
2. Introduction
3. Experimental populations
3.1. Advanced populations
3.2. Genotypic distributions
4. Statistical model of interval mapping for QTL
4.1. Statistical model
4.2. Conditional probabilities
5. Likelihood of the statistical model
6. Score test statistics
6.1. Score functions
6.2. Score test statistics
6.3. Asymptotic forms of score test statistics
7. Gaussian stochastic process
7.1. Covariance between test statistics
7.2. Covariance between trait means
8. Simulating the null distribution
9. Real example and simulation studies
10. Discussion
11. Acknowledgements
12. Appendix
13. References

1. ABSTRACT

Issues in determining the threshold values of QTL mapping are often investigated for the backcross and F2 populations with relatively simple genome structures so far. The investigations of these issues in the progeny populations after F2 (advanced populations) with relatively more complicated genomes are generally inadequate. As these advanced populations have been well implemented in QTL mapping, it is important to address these issues for them in more details. Due to an increasing number of meiosis cycle, the genomes of the advanced populations can be very different from the backcross and F2 genomes. Therefore, special devices that consider the specific genome structures present in the advanced populations are required to resolve these issues.  By considering the differences in genome structure between populations, we formulate more general score test statistics and Gaussian processes to evaluate their threshold values. In general, we found that, given a significance level and a genome size, threshold values for QTL detection are higher in the denser marker maps and in the more advanced populations. Simulations were performed to validate our approach.