Hello, I am a graduate student and fairly new to RNA-Seq. In my study I have a 2x2 design with each group containing 4 bio reps and each bio rep has 3 technical replicates (same library prep, but each sample is processed in 3 different lanes). After studying my frequency count data I have found that the count data in lane 1 is very different from lane 2 and 3 for most to all bio samples. I plan to use EdgeR for DE analysis and I am wondering if it is OK to sum count totals from just lanes 2 and 3. All documentation I have read states that technical variation is small so you should sum all tech reps, but in this case it seems I have noticed greater variation in one lane and therefore it does not make sense to include it. I have not found any reason why you can not just sum 2 out of 3 technical replicates if you know that one technical replicate has much higher variation than the other two. Please let me know if you have any evidence for this. Thank you very much in advance. -- Neha Mehta --------------------------------------- PhD Candidate, Neuroscience Northwestern University ph. (412) 874-6342 e. nsmehta@u.northwestern.edu [[alternative HTML version deleted]]

Hi there, No, there is absolutely no problem with dropping a lane of data if you believe that lane to have technical issues. I would recommend that you use the plotMDS function to verify that the counts in the bad lane are indeed different from the other two lanes. -Ryan On Fri 21 Mar 2014 01:45:51 PM PDT, Neha Mehta wrote: > Hello, > > I am a graduate student and fairly new to RNA-Seq. In my study I have a 2x2 > design with each group containing 4 bio reps and each bio rep has 3 > technical replicates (same library prep, but each sample is processed in 3 > different lanes). After studying my frequency count data I have found that > the count data in lane 1 is very different from lane 2 and 3 for most to > all bio samples. I plan to use EdgeR for DE analysis and I am wondering if > it is OK to sum count totals from just lanes 2 and 3. All documentation I > have read states that technical variation is small so you should sum all > tech reps, but in this case it seems I have noticed greater variation in > one lane and therefore it does not make sense to include it. I have not > found any reason why you can not just sum 2 out of 3 technical replicates > if you know that one technical replicate has much higher variation than the > other two. Please let me know if you have any evidence for this. > > Thank you very much in advance. >