I obtained a significant association in a case control study for a haplotype and also I did a statistical test for each SNP individually in that haplotype.
I wonder if there is a statistical test for determining that the significance obtained in the haplotype test is better than the one obtained within the individual SNP statistical test.
Can anybody give me a clue?
Dear Dietmar, if you want to test the effect of haplotype and SNP on a variable, and your data are of continuous scale and normal. You can simply apply a T-test between them. But if you have several haplotypes and SNPs, and aim to figure out the interaction effects of them on a dependent variable you'd better apply a two-way ANOVA or RSM ( response surface regression model) method.
As it was case-conrtol study, so outcome was binary.
One of way -to fit logistic regressions 1. Outcome~f(haplotype) and 2. Outcome~f(SNP)
Then models can be compared based on LR or Chi-square tests.
Thank you both Ehsan and Olga for your answers, I will try both methods.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis