I would like to use the publically available results from CARDIoGRAMplusC4D but can't work out which allele is the effect allele. The data has fields for "reference_allele", "other_allele", and "eaf". But eaf, which I assume stands for effect allele frequency, seems to correspond to either the reference allele or the other allele. I worked this out using dbSNP and eaf to work out the effect allele (excluding palindromic SNPs and eaf close to 0.5). Has anyone else got experience working with this dataset?
http://www.cardiogramplusc4d.org/downloads/
I think I managed to answer my own question.
I compared 23 SNPs from the main paper for cardiogramplusc4d (Deloukas et al 2013) to the results file for cardiogramplusc4d . Basically, I found that the reference allele in cardiogramplusc4d always corresponded to the effect allele in that paper. However, the eaf column in cardiogramplusc4d was inconsistent with the effect allele frequency in Deloukas et al. Thus, it looks like the eaf column in the cardiogramplusc4d text file is unreliable.
There is inconsistency between C4D, CARDIoGRAM and CARIoGRAMplusC4D in terms of what is called the effect allele but I have not found any occasions where the sign of the beta has not been swapped appropriately.
I think the data are OK; just don't assume they match the earlier analyses.
The problem was resolved. Apparently the effect allele frequency column was incorrect. This has now been fixed.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools