natsuhiko/rasqual

Understanding Effect Sizes

arielmadr opened this issue · 1 comments

arielmadr commented

Dear Natsuhiko,

I am going through the results of Rasqual and I found some QTLs with high effect sizes (0.8) and good Q-values ( 8.33E-13 ), but when I look at the coverage across the three groups I see no difference between my three groups ( homozygous reference, heterozygous, homozygous alternative). Could this be false positives or am I missing something ( should I check the values of other columns like Delta or Phi)?

Thanks a lot,
Ariel

natsuhiko commented

Hi Ariel,

Although I haven't looked at your data and I cannot conclude, it could be a false positive. You should have a look at allele specific counts at each feature SNP as well. It might be due to a SNP with low coverage or some mapping error (that RASQUAL is not able to capture well). You can also fit the model without allele specific counts (--population-only option) to make sure there is no QTL signal in the total read counts.

Best regards,
Natsuhiko