Comparing metrics of two exome: how to interpret higher median coverage but lower mean target coverage?

Question

[pegeot.henri]

Hello there, so I am analyzing metrics from two kit of exome sequencing and I have the following:

For 100M reads for both exome:

Qualimap: median coverage:

• Exome A = 150X
• Exome B = 110X

Picard HS Metrics mean target coverage:

• Exome A = 75X
• Exome B = 85X

How do you interpret that? Uniformity of coverage is better for exome B? Which metric could I use to investigate further?

[Jean-Karim Heriche]

Doesn't Picard HS also output median coverage ? From a statistical perspective, you can't compare mean and median without making the assumption that for the underlying distributions the mean and the median are equal. Also mean and median are only useful for comparison if the underlying distributions are unimodal.

[pegeot.henri]

Doesn't Picard HS also output median coverage ?

Yes and it is different from qualimap and I have no idea why, for the same data: Qualimap median coverage = 110X ; Picard Median coverage = 80X I know picard does not consider duplicated read, I cannot find out if Qualimap does it...

Thank you, I know now that I have to look distribution to have a better understanding of my data

Answer 1

[trausch]

For your downstream analyses you probably want to know how many exons (targets) are above a certain coverage threshold. In other words, you want to plot the fraction of targets >= X and then iterate X from 1x, 2x, ..., 50x coverage. In addition, you want to look at the on-target rate, i.e., how efficient your enrichment worked. Among other tools, our method Alfred is generating these metrics.

alfred qc -r genom.fa -j out.json.gz -o out.tsv.gz -b exome.targets.bed input.bam

You can view an example here, just select the exome data set from the reference data set collection.