Multi dataset for CLI

[MarieLataretu]

Hi folks,

I recently came across the new web feature for multi-dataset usage and was wondering what a good equivalent would be using the CLI. (I didn't find the (pre-)alignment implementation in the code, maybe you could point me to that.)

Also, I was wondering if running all input samples (e.g., all Influenza) against all available (Influenza) datasets has drawbacks, besides wasting resources.

Best
Marie

[ivan-aksamentov]

Hi @MarieLataretu,

Try nextclade sort command. It allows you to split your fasta sequences by their detected dataset. It uses the same algo as Nextclade Web when it suggests one or multiple datasets given a fasta file. After that, you could handle each subset independently with nextclade run. This could be automated with a bash script (and e.g. GNU parallel) or Snakemake etc.

Here is an example of Snakemake pipeline which sorts RSV to A and B and then processes them separately:

https://github.com/nextstrain/rsv/blob/170f99c2784c35beae52845cd8d84916196f8870/ingest/workflow/snakemake_rules/sort.smk#L22

Type nextclade sort --help for more details.

[MarieLataretu]

Hi @ivan-aksamentov , excellent, thanks a lot! We'll look into nextclade sort 👍