IUPAC degenerate nucleotide codes accepted as input?

[RhysCAllen]

Description of bug

Greetings SPAdes team,

Does spades.py parse input files with degenerate IUPAC DNA codes such as R,Y,K,M? Or does it process ACGTN inputs only?

I noticed unexpected behavior in running BayesHammer, and I'm unable to find spades documentation on degenerate nucleotide symbol inputs.

Specifically, with an input read using degenerate symbol R, such as
@S0R13002195/1
CATTTCTTGGAAAAATATATTCATTTC R GTGAAATCTAAATGTTGAGTTCGCGTTTCACAAGA

BayesHammer error correction returns a non-IUPAC symbol:

@S0R13002195/1
CATTTCTTGGAAAAATATATTCATTTC ? GTGAAATCTAAATGTTGAGTTCGCGTTTCACAAGA

AlphabetValidator: invalid sequence character: 'ÿ' reported by fq, a FASTQ linter utility.

Using grep, the actual symbol returned by BayesHammer for this ambiguous nucleotide is neither ? nor ÿ. AWK indicates the character may be invalid multibyte data.

Since the FASTA standard includes IUPAC definitions for degenerate DNA bases, but BayesHammer is returning a non-standard symbol, I'm uncertain whether this is a symptom of a bug, or this is expected behavior (e.g. SPAdes requires AGTCN-only input files).

Thank you for any additional info you could provide. If available, please share the relevant section of documentation / publication.

spades.log

NA

params.txt

NA

SPAdes version

4.2.0

Operating System

Linux-6.8.0-90-generic-x86_64-with-glibc2.39

Python Version

3.12.13

Method of SPAdes installation

binaries

No errors reported in spades.log

• Yes

[asl]

SPAdes and BayesHammer do expect ACGT input. Everything else should be automatically normalized just to A for the sake of clarity. Not sure what was the problem in this particular case though...

[RhysCAllen]

Interesting. Thank you for the reply! So, just to make sure I understand, the expected behavior is that BayesHammer gracefully (and silently) changes any degenerate input symbols such as R,W, or even N, all of those would automatically be changed to adenine (A)?

[asl]

(and silently) changes any degenerate input symbols such as R,W, or even N, all of those would automatically be changed to adenine (A)?

Correct. This is the intention. Not sure what happened in your particular case though and why this did not work.

[RhysCAllen]

I'm not sure either, though it is reproducible, and occurs across multiple different samples (the CAMI plant rhizosphere samples).

Could it be related to fragmented memory, or the mimalloc error?

Is this adenine substitution behavior in the spades documentation or publications somewhere? I thought I had read the docs pretty thoroughly, but would be great if there was an additional undiscovered stash of info for me to check out.

Thanks again!