Hi,
I am unsure how to deal with this case to go about analysing RNA-seq data.
Suppose that you have a control and treatment setup with 4 biological replicates each. However, two in control and two in treatment were pooled together and paired-end sequenced and the other 4 were single-end sequenced. That is:
Condition Sample_No Sequence_Type
Control 1,2 paired-end
Control 3,4 single-end
Treatment 1,2 paired-end
Treatment 3,4 single-end
Now, suppose I'd want to perform a differential gene expression analysis, how would you take care of the difference in the reads (due to PE and SE) within conditions?
i) You map the reads as such - PE as PE and SE as SE libraries. You count the total number of reads that fall under each sample. You then normalize using edgeR's TMM method for difference in the library size and then perform the differential expression analysis.
ii) You map the reads as such - PE as PE and SE as SE libraries. You count the total number of reads first in pair (meaningful for PE samples) that fall under each sample. You then normalize using edgeR's TMM method for difference in the library size and then perform the differential expression analysis.
iii) You discard the second pair altogether and treat them as two conditions with 8 SE libraries.
Understanding the inherent mess-up in the experimental setup and the possibility of bias and ...