how do you describe the sequencing depth of your sequencing data

[haiwan]

Hello everyone,

i am a junior research. I have worked on the deep sequencing for human miRNA profiling in human plasma samples. I recently prepared the manuscript relating to miRNA profiling obtained by Solexa sequencing.

During the preparation, i have encounter an issue about the sequencing depth. I already looked up the literature. Though i understood what the depth is, i am still confused with this concept.

The common description of coverage is shown as 30X, for example.

So, my question is like that:
How do you depict the coverage of miRNA sequencing data? Is it according to the equation referring to C=LN/G was used to calculate the coverage, where C stands for coverage, G means the haploid genome length, L is the read length, N is the number of reads?


thanks a lot for help.

sincerely,

[dpryan]

Just use the number of reads and percentage uniquely aligned. Coverage only makes sense in the context of DNA.

[haiwan]

thank you for your reply.
i am still confused. how do you specify the coverage if the reviewer ask you this question?

with best,

[dpryan]

You tell them that what they're asking is meaningless and they should learn to stop making stupid requests. Of course, you don't phrase it like that.

Something of the form: "We thank the reviewer for his/her helpful remarks. Regarding reporting coverage, there is unfortunately no straight-forward way to do so with miRNA profiling (or RNAseq in general). A single coverage metric is quite meaningful in many high-throughput sequencing applications, since relatively uniform coverage is expected. Common forms of RNAseq, however, do not and should not produce even remotely uniform coverage of known or unknown transcripts. This is due to the vast concentration range that various RNAs can assume. A coverage metric, then, can be quite misleading in this context, particularly since many transcripts are simply not expressed in a given context."

Something along those lines will work. The thing is just that the reviewer doesn't know much about NGS, but has just read enough papers to know a bit of the lingo.

[GenoMax]

I wonder if the choice of the word "coverage" by haiwan was not precise enough in the original post.

Perhaps the metric should have been "depth" observed in the actual data collected for this experiment.

[dpryan]

The two are often synonyms, so I would try to avoid them both.

[haiwan]

thank you for reply.

it is more and more confusing. I understood there is a bit difference existing between coverage and depth.
if yes,
How do you know depth for your data? Can you calculate according to some sort of the equation? The technician would know the sequencing depth because she/he sets the parameters for the specific aims?

I wonder if the choice of the word "coverage" by haiwan was not precise enough in the original post.

Perhaps the metric should have been "depth" observed in the actual data collected for this experiment.

[GenoMax]

Some additional points of view.

http://www.biostars.org/p/638/

http://www.biostars.org/p/6571/

http://www.researchgate.net/post/What_is_the_difference_between_sequencing_coverage_and_depth

Based on the data at hand "how many times a particular base has been covered by a sequence read" is a real observable entity. Whether to call it depth/coverage or something else (Picard HsMetrics uses the X coverage for a target base) is open for discussion.

[dpryan]

@GenoMax, nice examples! Of course, they all apply to DNA sequencing. I object to using "coverage" and "depth" here only because I don't think they're useful in the context of miRNA profiling (or RNAseq in general).

[CofactorGenomics_Natalie]

Haiwan-

Thanks for initiating a thoughtful discussion. At Cofactor Genomics, we generally avoid using the terminology "depth" and "coverage" for RNA-sequencing as well. Reporting the number of reads sequenced is an acceptable practice.

The other option you have is to perform a saturation analysis, where you evaluate the number of known miRNA that you've sequenced in your experiment. Reporting this add additional insight to your data. If you have questions regarding this type of analysis, please let me know and I'd be happy to discuss.

[haiwan]

Thank you very much indeed.

i learned more from all of you. It is quite complex for sequencing data. I think it is very important from start to learn.
I am going to send samples for 16s sequencing. So, it would be quite different from miRNA sequencing? Another story again?

[dpryan]

Well, for the whole "reporting depth/coverage" thing, I would say that's the same for 16s profiling. The amount of sequence you need will depend on the complexity of the underlying sample. Consider the saturation analysis suggested by Natalie@CofactorGenomics, that's an excellent way to go about this (and should shut-up any reviewer and put your mind at ease regarding having sufficient signal).