Incorrect placement of ORF9b amino acid substitution adjacent to a deletion (lineage XFC.1)

[LucvZon]

Nextclade appeared to misplace an amino acid (AA) substitution when it occurs immediately before a deletion.

In our analysis of a SARS-CoV-2 sequence (lineage XFC.1), we observed a nucleotide substitution G28360T which should cause an M25I AA change in ORF9b. This substitution is immediately followed by a 9-base deletion from 28361-28370.

Instead of reporting M25I, Nextclade reports a deletion at AA position 25 and incorrectly moves the resulting Isoleucine (I) to AA position 28, which is part of the deleted region.

Here is the relevant snippet from aaChangesGroups in nextclade.json:

[
  {
    "cdsName": "ORF9b",
    "pos": 25,
    "refAa": "M",
    "qryAa": "-", // BUG: Incorrectly reported as a deletion
    "nucPos": 28358,
    "refTriplet": "ATG",
    "qryTriplet": "ATT" // CONTRADICTION: Query triplet is ATT (I), not a deletion
  },
  {
    "cdsName": "ORF9b",
    "pos": 26,
    "refAa": "E",
    "qryAa": "-",
    "nucPos": 28361,
    "refTriplet": "GAG",
    "qryTriplet": "---"
  },
  {
    "cdsName": "ORF9b",
    "pos": 27,
    "refAa": "N",
    "qryAa": "-",
    "nucPos": 28364,
    "refTriplet": "AAC",
    "qryTriplet": "---"
  },
  {
    "cdsName": "ORF9b",
    "pos": 28,
    "refAa": "A",
    "qryAa": "I", // BUG: The 'I' from pos 25 is misplaced here
    "nucPos": 28367,
    "refTriplet": "GCA",
    "qryTriplet": "---" // CONTRADICTION: Query triplet is a deletion, not 'I'
  }
]

The nucleotide G28360T mutation can be seen in the nextclade.aligned.fasta result:

Steps to Reproduce

1. Analyze a SARS-CoV-2 sequence that has the G28360T substitution and the 28361-28370 deletion. This issue has been reproduced with the following public GISAID accessions from lineage XFC.1:

• EPI_ISL_20058200
• EPI_ISL_20066338

2. Use Nextclade v3.16.0 with nextstrain/sars-cov-2/wuhan-hu-1/orfs.
3. Inspect the the aaChangesGroups array for the CDS ORF9b in the nextclade.json output.
4. Observe the nucleotide changes in the nextclade.aligned.fasta to confirm the G->T at ref position 28360 and the deletion at 28361.

I've also attached our input sequence file (change extension to .fasta) and the full nextclade.json result to help with debugging.

input_sequence.txt

nextclade.json

[ivan-aksamentov]

Hi @LucvZon,

Thanks for the detailed report. We'll investigate.

This is what I see for the provided input_sequence.txt in Nextclade Web.

[rneher]

hi @LucvZon

thanks for reporting this. The underlying issue here is that nucleotide sequences and amino acids are aligned separately (though we try to place gaps in the nucleotide alignment at codon boundaries). In this case, the combination of the nucleotide mutation and the 9base deletion results in the nucleotide sequence to be placed at codon position 26, and the corresponding amino acid at position 29 (for SC2, nextclade prefers to put the gap on the left). This results in the inconsistency you spotted.

The nextclade.json file contains a lot of data, part of which is really only intended to be used for the visualization that Ivan showed. This is meant to give an idea of the context around a change in the sequence.

[LucvZon]

Hi @rneher

Thank you for the explanation. I suppose we just have to be aware of inconsistencies every now and then when it comes to the visualizations.