report median read length for fastqc

[jchorl]

Uses median read length for fastqc, instead of mean. Note that this is a breaking change. Another option is to add an extra column, to keep both.

The median is way more suitable for sequencing reads as it doesn't assume the length is normally distributed.

There are issues on fastp to output more stats, but they don't seem to have much traction. I suppose I could also try to PR them, but they seem less responsive.

You can see a subtle difference in lengths on the testdata repo:
Old:

New:

• This comment contains a description of changes (with reason)
• CHANGELOG.md has been updated

[ewels]

Yeah, I wrote this code back before long reads were really a thing and it was more of a fair assumption 😅

I'm a bit nervous about the breaking change as this is by far the most used module. As such, I'd prefer to keep the previous average and add this in as an additional value if possible. At least then the same value as before will be present in the data exports for downstream analysis.

Also, I'd like the column title to be slightly different, eg. Median Read Length so that there's a visible difference between reports.

[jchorl]

Done! There is now an optional median column (hidden by default):

[ewels]

Suggested change

	hide_seq_length = False if max(avg_seq_lengths) - min(avg_seq_lengths) > 10 else True
	hide_seq_length = False if max(median_seq_lengths) - min(median_seq_lengths) > 10 else True

[jchorl]

This should really just be:

hide_seq_length = max(median_seq_lengths) - min(median_seq_lengths) <= 10

[jchorl]

Suggestions incorporated and tested:

Will cut an issue shortly on fastqc

[ewels]

Looks great. Thanks for the tweaks!