add some AVITI data to the fastq-test-suite

[colinbrislawn]

This is the new Illumina competitor from Element Biosciences @Elembio

People are using it amplicons: qiime2 forums xref, benjjneb/dada2#1893, company website

The q-scores are higher than Illumina's but idk what that means yet

I looked for public test data and couldn't find any

[frederic-mahe]

thanks @colinbrislawn

related issues: torognes/vsearch#522 and torognes/vsearch#564

As of now, vsearch can handle fastq quality values up to 41 by default and up to 93 with the proper option (--fastq_qmax 93). That's the end of the ASCII table when using an offset of 33 (93 + 33 = 126 = ~, the last printable ASCII symbol).

We could:

• change nothing (fatal error when Q > 41),
• accept Q values between 41 and 93 with a warning,
• set 93 as the new default upper limit for Q values

[colinbrislawn]

Well, there's no fastq spec. 'standards' https://xkcd.com/927/

For my own notes, Wikipedia has a surprisingly good section on the Phred offset debacle:
https://en.wikipedia.org/wiki/FASTQ_format#Encoding

I would love to have a good way to autodetect offset and help users fix mistakes, but I'm not sure that's possible.

Perhaps fastq 'correctness' should come upstream and we simply trust the scores as written.

set 93 as the new default upper limit for Q values