search

natsuhiko / rasqual

Robust Allele Specific Quantification and quality controL
37 stars 20 forks source link

RASQUAL with biological replicates #28

Open agolicz opened 5 years ago

agolicz commented 5 years ago

Hi, I was wondering. What is the best way to incorporate biological replicates in RASQUAL study design? I have RNA-Seq/ATAC-Seq data for 100 individuals (2 biological replicates per individual) The scenarios I can think of:

  1. Perform the analysis on two sets of replicates and compare the results
  2. Include replicates in RASQUAL study design (not sure what is the best approach here though) I suppose in eQTL scenario all samples with the same allele at the locus are considered replicates, so I can probably just lump all the data together, but I wonder if there is a better way.

What do you do in situations like that?

All the best, Agnieszka

natsuhiko commented 5 years ago

Hi Agnieszka,

Unfortunately RASQUAL is not able to take advantage of replicated data structure. Naively I would merge the replicates into one and map QTLs to maximise the coverage depth of feature SNPs. Otherwise I would keep the pair of replicates and permute samples to construct an empirical null P-value distribution from which we can compute FDR threshold. Unfortunately, the permutation function to keep the replicated data structure is not implemented and you need to permute data by yourself.

Best regards, Natsuhiko

agolicz commented 5 years ago

Thanks very much! I will try that. Do you happen to know of any QTL/eQTL software that does have native support for replicates?

Agnieszka

natsuhiko commented 5 years ago

Hi Agnieszka,

I have never heard any software that can handle replicates.

Best regards, Natsuhiko