I would like to perform QTL analysis on a list of BAM files that I have containing the locations of histone mark reads. Ultimately my aim is to perform Mendellian randomisation, with these hQTLs as the exposure, against a different phenotype as the outcome. The outcome effects are expressed as beta coefficients, SE and p-values from standard linear modelling.
However, the most popular QTL mapping softwares used for this type of analysis (WASP and RASQUAL) have effect sizes that are not directly comparable to the output of a linear model.
NOTE: I have been told that one could potentially convert RASQUAL output to beta, SE and p-values by doing the following:
beta = log(Pi/(1-Pi))
SE = sqrt(Chi-sq/Beta^2)
Above the beta
is being approximated by the Pi
effect size and subsequently (in theory) one should be able to compute SE from beta and Chi-square
value (11th column) so that the significance level of Wald statistic is identical to the Chi-square statistic.
However this is not standard practice....
Hence I am wondering if there is a software that one could use to model e.g. chip-seq peak height against genotype via a standard linear model in order to have results that could be used in a MR analysis against a typical GWAS output.
To summarise: I am wondering if anyone can either tell me if the above code is sufficient to make RASQUAL output comparable to standard linear model output. If not could someone point me in the direction of a workflow that allows for the modelling of histone marks in the way that I require.