I am wondering if anyone has looked at the effect of sequencing depth on the ability to detect differences in RNA-Seq count data? There are many useful references for RNA-Seq DE analysis, mostly addressing appropriate distributional models for count data, normalization and appropriate statistical design of experiments. I haven't seen a discussion of sequencing depth, though.
In our case, we have been analyzing human tissue RNA samples on both GAII and HiSeq platforms and have already generated a fair amount of data. In future studies we will have the option to run one sample per lane or multiplex up to some level but one question for us is what hit in ability to detect differences do we take as we reduce sequencing depth?
Some analyses of our current data and further experiments suggest themselves to us. It would be nice to hear from others who may have already thought about this problem or can point me to publications that might address it.
-JMorlan
In our case, we have been analyzing human tissue RNA samples on both GAII and HiSeq platforms and have already generated a fair amount of data. In future studies we will have the option to run one sample per lane or multiplex up to some level but one question for us is what hit in ability to detect differences do we take as we reduce sequencing depth?
Some analyses of our current data and further experiments suggest themselves to us. It would be nice to hear from others who may have already thought about this problem or can point me to publications that might address it.
-JMorlan