What are the disadvantages of increasing the sequencing depth of RNA-seq experiment?

Question

[mrj]

I am designing an RNA-seq experiment to study a region of the mouse brain. I am particularly interested in the Orexin system (set of rarely expressed genes). I have the option to sequence at a depth of 60 Million Paired-end reads or 100 Million paired-end reads. From what I have read, I realize that iIshould increase my sequencing depth. (I have decided to have 3 biological replicates from cases and 3 biological replicates from controls.)

If I increase the sequencing depth (say 100 Million Paired-End), will there be any disadvantages that I did not foresee? can this become computationally cumbersome to analyze? I am using a cluster of 64 nodes.

I would really appreciate it if I can get your experienced opinion about this.

[ATpoint]

Seconding the below answer. Sure a good depth is important for lowly-expressed genes but these have low statistical power due to low counts. If you can then do more replicates to gain statistical power, see http://m.rnajournal.cshlp.org/content/22/6/839.full The gain of power is especially notable if you do like four or five rather than two or three replicates. 60M is enough in most cases.

[mrj]

Thank you ATpoint. Thanks for the reference and suggestion.

Answer 1

[swbarnes2]

Fastq length doesn't matter like genome size does. I would give some thought to whether you'd be better off with more replicates, even if they are at lower depth.

[mrj]

Thank you swbarnes2