Hi, I'm quite new to RNA-seq experiments. Recently, I just got the results RNA-seq results for 4 groups of samples, and in each group there are 3 biological replicates. After mapping and quantification (into TPM), I normalized the TPMs according to the size factor (DESeq). Overall, a majority of data make sense to me and are quite consistent within each group (among three biological replicates), but I found a subset of transcripts whose TPM still vary a lot among three biological replicates. For example, in control group, TPM of transcript A in replicate 1 is 20, in replicate 2 is 25 and in replicate 3 is 10. I estimated there are about 5000 transcripts in each group exhibit such huge variation, which make the correlation results look not quite decent.
I wonder, can I just do a outlier test (such as Grubb's test) to remove the transcripts with outlier within three replicates? and after cleaning all the outlier transcripts in each group, I will compare the transcript level shared in four groups.
Thanks a lot.