Dear BioStars users,
I had a question I wondered if people wouldn't mind advising on. We have some ChIP-seq datasets from our lab we would like to combine with another, published ChIP-seq dataset from a different lab. When processing the data we trimmed our data to 70 nucleotides before continuing with the data analysis. However, when coming across this extra dataset from a different lab that we would like to combine our data with it seems the read quality is really poor at 51 nucleotides, so I am thinking of trimming the reads to 50 nucleotides.
My question is whether I would need to apply this same trimming at 50 nucleotides that I will use in this extra dataset to the existing data we have (i.e. remove the data obtained after a 70 nucleotide trim)? Would this not really make much difference if the two separate datasets were trimmed at different lengths, or would it drastically impact on any data analysis we perform down the line?
I really appreciate any advice you can offer. Thanks, Oliver
Hi, I would anticipate that from all the potential batch effects that ChIP experiments suffer from the read length is the least of your problem, but yes in general one should remove effects like mappability bias if possible. Still, the technical batch effect between indepedent experiments, especially experiments that involve tricks components such as antibodies, is probably notable.