5 Mill sounds few for me but I am not an expert and I am do not work with mmamalians genomes.

Any case, more than focus on the depth try to get as much  biological replicates as possible. To identify DEG, It is better 3 RNA seq of 5 Mill, than one experiment of 15 Mill.

I concur with Pedro. You need to focus on biological replicates rather than sequencing depth when considering differential expression.

See these papers:

http://bioinformatics.oxfordjournals.org/content/30/3/301.full

http://rnajournal.cshlp.org/content/early/2016/03/28/rna.053959.115

http://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-14-91

Also, a MiSeq is not suitable for mammalian RNA-seq. You'd have to perform many runs to get enough data to do the analysis, all of which has the potential to introduce confounding batch effects.

HiSeq or NextSeq data will be cheaper and give better data.

Hi,

can you clarify if you want to quantify only the therapeutic product or all the transcriptome? Relating this on your question, think that less is your sequencing depth less is your power to resolve not abundant transcripts. You said that the cell is design to express those therapeutic products at high levels, but how is in proportion to high expressed cellular transcripts?

Assuming you wants only those products, have you got a reference to align to reads or are you going to assemble them? In the first case, I can not see any advantage for the 2x300bp. In the second case, they wold be very useful for me as well as an higher sequencing depth.

Sorry, these are more consideration than a practical answer to your question but they may be useful. Finally, I surely would take Pedro suggestion to push on the number of experimental replicates, so if you have to discard the sequence for a sample for any reason won't be a major issue. 

Thank you Pedro, Christian and Luca for taking the time to answer! All advice is very much appreciated. I'll try to address all responses in this answer. 

Very interesting re the biological replicates. I should have said that we are planning to run a total of 12 samples - so my final comparison will be 6 versus 6 clones which were isolated from one cell line. 

I am also leaning towards 15M reads rather than 5M

Thanks also for the advice re instrumentation..... I will keep this in mind for future experiments 

We are not looking to quantify the product at all ... rather the rest of the transcriptome. I'll run some PCRs to compare to housekeeping genes though .. this is a good suggestion. 

Many thanks again

Clair

perhaps reading about quantseq will give some nw info for gene expression related  rnaseq experiment ( easier analysis  , lower costs)

https://www.lexogen.com/

I will add my two cents here - usually the sequencing centre you are using should give you a guideline, but a rough rule of thumb for human RNA-Seq for differential expression is about 50 million reads per sample. You can run around 5 samples per lane, with will cost you around on average 250 GBP per sample.

This is off the top of my head from what I can recall, have a look at the sequencing centre protocols.

The long reads produced by MiSeq are not really suitable for differential expression analysis in mammalian cells. Long reads are useful for de novo assembly and transcriptome analysis, e.g. looking at alternative splicing. However the read length will not much affect your statistical power of detecting differential expression (with the exception of some low-complexity genes). DEG analysis pipelines often use raw genecounts as input (e.g. DESeq2) and in this scenario, a long read counts the same as a short one, if it is uniquely mapped. As a rule of thumb, for a typical DEG experiment with mouse or human cells I would pick triplicates at ca. 20 M reads each, for example single-end 75 bp reads on a NextSeq flow cell. This read length is sufficient to map most reads uniquely.

Hi all,

Thank you all for your excellent advice! I decided to cancel the MISeq run and went with the HiSeq instrument instead. We sequenced extra replicates at a high depth & it has worked really well. Im delighted with the results and after reading around some more am sure that the experiment would have failed without your excellent advice re switching instruments.

Thanks so much again

Clair

Even if it's a little bit late, there is a nice article here:

RNA-seq differential expression studies: more sequence or more replication?

Yuwen Liu, Jie Zhou and Kevin P. White

Article RNA-seq differential expression studies: More sequence or mo...