Generalized composite interval mapping offers improved efficiency in the analysis of loci influencing non-normal continuous traits
In genetic studies, most Quantitative Trait Loci (QTL) mapping methods presuppose that the continuous trait of interest follows a normal (Gaussian) distribution. However, many economically important traits of agricultural crops have a non-normal distribution. Composite interval mapping (CIM) has been successfully applied to the detection of QTL in animal and plant breeding. In this study we report a generalized CIM (GCIM) method that permits QTL analysis of non-normally distributed variables. GCIM was based on the classic Generalized Linear Model method. We applied the GCIM method to a F2 population with co-dominant molecular markers and the existence of a QTL controlling a trait with Gamma distribution. Computer simulations indicated that the GCIM method has superior performance in its ability to map QTL, compared with CIM. QTL position differed by 5 cM and was located at different marker intervals. The Likelihood Ratio Test values ranged from 52 (GCIM) to 76 (CIM). Thus, wrongly assuming CIM may overestimate the effect of the QTL by about 47%. The usage of GCIM methodology can offer improved efficiency in the analysis of QTLs controlling continuous traits of non-Gaussian distribution.
| Main Authors: | , , , |
|---|---|
| Format: | Digital revista |
| Language: | English |
| Published: |
Pontificia Universidad Católica de Chile. Facultad de Agronomía e Ingeniería Forestal
2010
|
| Online Access: | http://www.scielo.cl/scielo.php?script=sci_arttext&pid=S0718-16202010000300007 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In genetic studies, most Quantitative Trait Loci (QTL) mapping methods presuppose that the continuous trait of interest follows a normal (Gaussian) distribution. However, many economically important traits of agricultural crops have a non-normal distribution. Composite interval mapping (CIM) has been successfully applied to the detection of QTL in animal and plant breeding. In this study we report a generalized CIM (GCIM) method that permits QTL analysis of non-normally distributed variables. GCIM was based on the classic Generalized Linear Model method. We applied the GCIM method to a F2 population with co-dominant molecular markers and the existence of a QTL controlling a trait with Gamma distribution. Computer simulations indicated that the GCIM method has superior performance in its ability to map QTL, compared with CIM. QTL position differed by 5 cM and was located at different marker intervals. The Likelihood Ratio Test values ranged from 52 (GCIM) to 76 (CIM). Thus, wrongly assuming CIM may overestimate the effect of the QTL by about 47%. The usage of GCIM methodology can offer improved efficiency in the analysis of QTLs controlling continuous traits of non-Gaussian distribution. |
|---|