Protein sequence validation

[simarilion]

Hi seqkit team,

I'd like to simply validate some protein sequences but the behaviour of seqkit is not as I'd intuitively expect.....

seqkit version -u
seqkit v2.9.0
Checking new version...
You are using the latest version of seqkit

echo -e ">seq\nMFKXXXXXQLRTNKZZZZZDRTFPAD" | seqkit seq --seq-type protein -v

seq
MFKXXXXXQLRTNKZZZZZDRTFPAD

I would expect some warnings about a protein sequence with "X" or "Z" rather than just returning the input sequence. Am I missing something?

Any advice would be gratefully received!

[shenwei356]

How about showing IDs with Z or X in the sequences

$ echo -e ">seq\nMFKXXXXXQLRTNKZZZZZDRTFPAD"  \
   | seqkit grep -s -P -i -p z -p x | seqkit seq -ni
seq

[simarilion]

Thanks for you fast reply Shen Wei!

Yes, that works - thanks! But I guess as a naive user I would expect a warning about such protein sequences as "Z" and "X" are not standard amino acids.

[shenwei356]

They are allowed in protein sequences 🥲 Someone asked to add Z a long time ago.

https://github.com/shenwei356/bio/blob/master/seq/alphabet.go#L34-L73

	Z	Glx	Glutamine or Glutamic acid [2]
	X	unknown amino acid

 2. http://www.dnabaser.com/articles/IUPAC%20ambiguity%20codes.html

[simarilion]

Hmmmm, I see the conundrum. Problem is for many sequence alignment and phylogenetic analysis programs "X" and "Z" residues are hard rejected. Maybe some kind of 'soft warning' would be good??

Nonetheless I greatly appreciate your helpful and fast responses :-)!

[shenwei356]

How about replacing Z with Q or E, and replacing X with any residue?

$ echo -e ">seq\nMFKXXXXXQLRTNKZZZZZDRTFPAD" \
    | seqkit replace -s -i -p Z -r E \
    | seqkit replace -s -i -p X -r M
>seq
MFKMMMMMQLRTNKEEEEEDRTFPAD

[simarilion]

In my case I can't replace residues with something else as these are supposed to be explicit datapoints - i.e. every position in the protein alignment should be explicitly defined (and not in the case of "X" whatever you want!).

My plan is to use seqkit for this very purpose; check that all sequences in an alignment do not have undefined residues. If they do then probably remove them from the analysis!