Inconsistent QC Results Between Nextclade CLI and Web Version

General
1

Hello everyone,

I’m working with a SARS-CoV-2 dataset and trying to filter sequences based on qc.overallStatus using Nextclade.

When I run Nextclade via the command line, all my sequences are labeled with “good” quality. However, when I use the web version of Nextclade on the same sequences (after updating the FASTA file), a significant number are marked as “bad” or “mediocre” quality.

I’m using the same reference dataset in both cases, Wuhan-Hu-1/2019 (MN908947), so I’m puzzled by the discrepancy in QC results.

Has anyone encountered this? Do you know what could explain the difference between the CLI and the web version?

Thank you in advance!

2

Hi @georges.hanna

Nextclade dev here.

There are several reasons why this could happen.

Could you please provide some more info on how you use CLI and Web exactly?

  • What version of Nextclade CLI are you using (nextclade --version)? What version of Web (shown in the bottom right corner of the page)? For reproducible results the versions should match.
  • Have you downloaded a fresh dataset for CLI using nextclade dataset get command? Web normally always uses the latest available dataset at the time you load the page, so if CLI uses an older version, then the results could differ. You can compare “Updated at“ in Web interface and .version.updatedAt in pathogen.json file of the dataset you use with CLI. The latest tag is 2025-06-09T15:42:38Z.
  • Do you pass any extra arguments to CLI?
  • I assume you are using our official deployment of Nextclade Web at https://clades.nextstrain.org?
  • Do you use any extra URL parameters when loading Nextclade Web? (in the URL bar of the browser - they could come from a saved bookmark or from a link on a third-party site)
  • It could be that Nextclade Web has retained some stale cached data. Try to 1) reload the page bypassing cache with Ctrl+F5 (Cmd+F5 on mac) or 2) to clear site data for clades.nextstrain.org in your browser or 3) to run in a fresh private/incognito window which have not ran Nextclade previously

If the issue persists, then this could be a bug in Nextclade. Could you please provide some details on how I can reproduce this on my computer? I would need some example sequence data (and other data you use, if any) which causes the problem to appear, as well as full command line for CLI and all URL params (if any) for Web. This would help me to reproduce the issue and then debug and hopefully fix.

1 Like
3

Hi Avan,

Thank you again for your help!

The issue has been resolved. After confirming that both the web and CLI versions were aligned, I realized I hadn’t been using the --input-pathogen-json argument with nextclade run. That was the reason all sequences were previously classified as good quality by default.

I appreciate your support, it was really helpful!

Georges