Hi,
I am performing an RNA-seq experiment to look at differential expression. The design is as follows:
2 populations x 3 biological replicates x 3 food treatments = 18 samples total
My interest is to make comparisons between food types within each population, and also to compare populations on the same food type. I was originally planning to use two HiSeq lanes, but am concerned about lane effects if I were to, for example, do one population on each lane (9 samples per lane)--any comparisons between populations would be confounded by possible lane effects. Unfortunately, I don't think I can multiplex more than 12 to a lane (according to the rep I am talking with), so I can't spread all 18 samples over two lanes. So, I am considering two options:
1. Do 12 samples in one lane and 6 in the other. One lane would have 2 biological replicates from each population, and the other lane would have the other biological replicate from each population. At least this would be unbiased because pops/treatments are spread the same say across lanes. My concerns: is 12 to many in a lane to get appropriate coverage for each sample? Obviously depth will vary considerably between the samples in the two lanes--will this be difficult to deal with when I do the analysis or is this easily accounted for in normalization?
2. Run three lanes, 6 in each lane. I don't see any problems with this except that I would like to avoid the extra cost with running another lane if at all possible.
I'm looking for any guidance or other ideas more experienced people may have. Thanks!
I am performing an RNA-seq experiment to look at differential expression. The design is as follows:
2 populations x 3 biological replicates x 3 food treatments = 18 samples total
My interest is to make comparisons between food types within each population, and also to compare populations on the same food type. I was originally planning to use two HiSeq lanes, but am concerned about lane effects if I were to, for example, do one population on each lane (9 samples per lane)--any comparisons between populations would be confounded by possible lane effects. Unfortunately, I don't think I can multiplex more than 12 to a lane (according to the rep I am talking with), so I can't spread all 18 samples over two lanes. So, I am considering two options:
1. Do 12 samples in one lane and 6 in the other. One lane would have 2 biological replicates from each population, and the other lane would have the other biological replicate from each population. At least this would be unbiased because pops/treatments are spread the same say across lanes. My concerns: is 12 to many in a lane to get appropriate coverage for each sample? Obviously depth will vary considerably between the samples in the two lanes--will this be difficult to deal with when I do the analysis or is this easily accounted for in normalization?
2. Run three lanes, 6 in each lane. I don't see any problems with this except that I would like to avoid the extra cost with running another lane if at all possible.
I'm looking for any guidance or other ideas more experienced people may have. Thanks!
Comment