N50 is not a median

[Sebastien-Raguideau]

Hello,

I am using sequali from quay.io/biocontainers/sequali:0.5.1--py310h4b81fae_0, so maybe a bit outdated since 6 month old and maybe this issue is fixed already.

I have a gripe with the reported values for N50 and other Ns, it seems that reported values are quantiles instead of N50 and such.

Just to be clear N50 should be a contig size, the contig size for which 50% of all nucleotide from the assembly are in smaller sized contigs. So for instance N50 of 9kb would mean that 50% of all nucleotides are in contigs smaller than 9kb. This is quite different from median and quantiles.

Best,
Seb

[rhpvorderman]

You are correct, these are quantiles. I got the naming wrong. I will fix this. Thanks very much for reporting!

[Sebastien-Raguideau]

Thx for quick answer.
Cool.
Unsure if that is on your mind but N50, still something that would be cool stat to have for long reads.

[rhpvorderman]

I will think about adding it, but currently I am unsure how the N50 value would translate to actionable QC things to do. The quantile values are a bit clearer translation in this respect when it comes to length filtering. Feel free to convince me otherwise! I am always open to suggestions.

[Sebastien-Raguideau]

I think I see your point. Hum, I suppose I just like to look at N50 to get a grasp of length distribution and it is a nice summary stat of your dataset. I think that your report help indeed with actionable QC, but also get an overview of the sequencing run? I mean I can run other tool to get that info, like seqkit stats, but what if I didn't have to?

[rhpvorderman]

I took a look: https://bioinf.shenwei.me/seqkit/usage/#stats N50 is the only thing missing from Sequali at the moment. Is that correct?

I can add N50 and N90 stats. That seems to be quite useful as a summary statistic.

[Sebastien-Raguideau]

Think so. Would be nice :)

[rhpvorderman]

@Sebastien-Raguideau The latest release should have fixed the quantile issue.

[Sebastien-Raguideau]

Hey, thanks a lot, will have a go at it.