We are doing an RNA-seq experiment comparing a wild type sample to a mutant. There are three biological reps of each, each tagged with a unique barcode.
These will be run on two lanes of an Illumina Hi-Seq. Our core averages around 65 million reads per lane.
I'm thinking that the ideal setup here is to multiplex all 6 samples on both lanes and combine the reads for each sample afterwards. In theory, this should eliminate any lane-lane variation between the samples as this will be spread across all 6 samples.
Or am I wrong in this? Is it better to multiplex 3 of all our wild type in one lane and 3 of our mutant in another. Or maybe a design of 2 WT 1 mutant in one lane and 2 mutant 1 WT in the other. I would think in either case we should average ~20 million reads per sample, but that the first is the ideal setup.
Any thoughts on this? I know many claim that the technical variation observed is minimal, but I like to eliminate as many confounding factors as possible.
These will be run on two lanes of an Illumina Hi-Seq. Our core averages around 65 million reads per lane.
I'm thinking that the ideal setup here is to multiplex all 6 samples on both lanes and combine the reads for each sample afterwards. In theory, this should eliminate any lane-lane variation between the samples as this will be spread across all 6 samples.
Or am I wrong in this? Is it better to multiplex 3 of all our wild type in one lane and 3 of our mutant in another. Or maybe a design of 2 WT 1 mutant in one lane and 2 mutant 1 WT in the other. I would think in either case we should average ~20 million reads per sample, but that the first is the ideal setup.
Any thoughts on this? I know many claim that the technical variation observed is minimal, but I like to eliminate as many confounding factors as possible.
Comment