Dear all, I have an analysis of a data set that I am hoping to get some input on. I have an experiment where I score a phenotype for a number of iso-genic lines. Two of the lines are reference iso-lines representative of the two extreme phenotypes (high and low). Both of these lines are on a common genetic background except for their second chromosomes. The effect on the phenotype thus resides on the second chromosomes. This has been formally established. Then I have eight different iso-lines that were scored for the same phenotype, four of which were scored similar to the high phenotype and four which were scored similar to the low phenotype. These eight lines all differ in their genetic background including their second chromosomes. Next, I have these same eight lines put on a common background (the same background as the high and low reference lines) and consequently only differ by their second chromosomes where I expect the effect on the phenotype resides. I wish to model the interaction of genetic background and second chromosomes on the phenotype and at the same time using the two reference lines with high and low scores as benchmarks for high and low scoring phenotypes. So, each line is compared to itself with and without common background, but also to the reference lines. First, I thought about constructing a glm with line and background as factors and then running a post hoc analysis to see the direction of effects within and between lines. This generally works. However, I am thinking this is a messy approach and I am not sure I am achieving the idea of using the reference lines as benchmarks for phenotype scores. Any thoughts and ideas are most welcome. Thanks in advance Allan Edelsparre
Analyzing iso-genic lines with different chromosomes on common genetic background
2 messages · Allan Edelsparre, Bob O'Hara
1 day later
You should just be able to run a single GLM, and then extract the different comparisons as contrasts. If you're comfortable with contrasts it shouldn't be messy. If you're not, I would consider fitting the same model twice, once with each reference line as the intercept. The model is formally the same, but the contrasts are to each reference line. Having written that, I suspect the first thing to do is to fit a model with one reference line as the intercept, and then plot the estimated contrasts to that. Hopefully it will informally tell you most of what you want to know. Bob
On 08/12/2024 17:25, Allan Edelsparre wrote:
Dear all, I have an analysis of a data set that I am hoping to get some input on. I have an experiment where I score a phenotype for a number of iso-genic lines. Two of the lines are reference iso-lines representative of the two extreme phenotypes (high and low). Both of these lines are on a common genetic background except for their second chromosomes. The effect on the phenotype thus resides on the second chromosomes. This has been formally established. Then I have eight different iso-lines that were scored for the same phenotype, four of which were scored similar to the high phenotype and four which were scored similar to the low phenotype. These eight lines all differ in their genetic background including their second chromosomes. Next, I have these same eight lines put on a common background (the same background as the high and low reference lines) and consequently only differ by their second chromosomes where I expect the effect on the phenotype resides. I wish to model the interaction o f genetic background and second chromosomes on the phenotype and at the same time using the two reference lines with high and low scores as benchmarks for high and low scoring phenotypes. So, each line is compared to itself with and without common background, but also to the reference lines. First, I thought about constructing a glm with line and background as factors and then running a post hoc analysis to see the direction of effects within and between lines. This generally works. However, I am thinking this is a messy approach and I am not sure I am achieving the idea of using the reference lines as benchmarks for phenotype scores. Any thoughts and ideas are most welcome. Thanks in advance Allan Edelsparre [[alternative HTML version deleted]]
_______________________________________________ R-sig-ecology mailing list R-sig-ecology at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
Prof. Bob O'Hara Dept. of Mathematical Sciences NTNU Trondheim, Norway