gnomaAD exome and genome AF differences

Question

[cocchi.e89]

Hi all,

maybe it is a silly question but in my university script to annotate variants I have a couple of options relatives to gnomAD filtering:

--gnomad-exome-af: is to specify a threshold and it will require variants to pass the threshold in all populations from "--gnomad-exome-pop"

--gnomad-genome-af: is to specify a threshold and it will require variants to pass the threshold in all populations from "--gnomad-genome-pop"

what is exactly the difference amongst these? Are the exome AF relatives only on patients that dis a whole exam sequencing while the "genome" refers to all samples (or only to those who had genome sequencing and not exam seq)

Thanks a lot in advance for any help!

[Ram]

I think the gnomad-exome dataset includes AF from exome sequencing datasets only, such as ExAC or ESP6500, whereas gnomad-genome would include AF by counting alleles from WGS datasets as well.

Answer 1

[Pierre Lindenbaum]

there are exome and genome data in gnomad.

There are distinct values for the frequences (AF) of each method because there are different arctefact/samples for each method. e.g:

http://gnomad.broadinstitute.org/variant/1-120457932-C-T exome AF: 0.00007992 genome AF : 0.00006368

[cocchi.e89]

Thank you Pierre! Would one would you recommend to choose in order to take only rare variants?

[Pierre Lindenbaum]

Would one would you recommend to choose in order to take only rare variants?

no, it depends of the nature of the disease.

[cocchi.e89]

Thanks Pierre, can you be more exhaustive about that? As example, if I should analyze IgAN nephropathy which one would you suggest and why?

Thanks a lot again

[Pierre Lindenbaum]

I'm not a clinician + I don't know that disease.

You should work in pair with an expert of the disease.

[diogo.pellegrina]

Wooow, I see this is super old, but to help the next people that come here:

The exome will be most usefull for diseases in general, as thats the part that will make the proteins and even ncRNAs. Be aware that you might loose some promoter regions, but if unless you intend to look specifically for promoter regions (or your disease is known to be affected by that), you might be better off with the exome only. This is because the complete genome data is a LOT larger, and it might skew completely your multiple testing adjustments and create many intergenomic positives that are mosly false positives.

The full data is most usefull for people interested in inheritance, linkage disequilibrium, populational genetics, etc... As the intergenomic variants have significantly less evolutionary pressure they become good markers to show the passage of time. Also if you are interested in measuring DNA damage/repair, conservation on those regions might be used as a control for the neighbouring genes.

[humzakhan340]

The point that genomes are much larger and more computationally intensive to work with is well-taken.

However, promoter information is not the only thing lost when going from genomes to exomes. It should be noted that many novel diseases are described where intronic variants actually affect expression or regulation of a gene (via splicing or transcription factors or what have you; see here: https://pubmed.ncbi.nlm.nih.gov/28359783/).

Additionally, I'm not exactly sure what you mean by multiple testing adjustments, etc. If you are analyzing a disease, the genetic material is often filtered for rare variants of interest anyways; I don't often encounter multiple testing adjustments in clinical genetics unless we're talking about GWAS.