Incorrect MIME-type for TSV-file inputs

[ebete]

Description of the bug

The MIME-types for TSV files in nextflow_schema.json are set to text/tsv instead of text/tab-separated-values (see: IANA tsv).

• https://github.com/nf-core/ampliseq/blob/2.12.0/nextflow_schema.json#L17
• https://github.com/nf-core/ampliseq/blob/2.12.0/nextflow_schema.json#L26
• https://github.com/nf-core/ampliseq/blob/2.12.0/nextflow_schema.json#L60

This causes issues within Seqera cloud, as TSV-files uploaded using the Datasets tab will have the text/tab-separated-values MIME-type, and will thus not be selectable as an input for the pipeline as it doesn't match text/tsv.

See Seqera platform docs regarding this feature:
https://docs.seqera.io/platform/24.3/data/datasets#use-a-dataset

Command used and terminal output

Relevant files

No response

System information

• Seqera cloud: 24.2.3_5387adb
• nf-core/ampliseq: 2.12.0

[pinin4fjords]

Actually the Platform incompatibility is more to do with the lack of pattern specification in the schema- could be fixed like #832.

[d4straub]

The problem is solved for --input in the dev branch. However, all params that point to files should have patterns to make them work, so I keep the issue open as reminder.

[ebete]

It looks like TSV datasets are still not selectable from the dropdown within Seqera Platform when using the dev branch. But I'm also not sure if the platform uses extensions for matching datasets, and instead opts for MIME-types. At least in the front-end, the extension is stripped from dataset names.

[pinin4fjords]

It looks like TSV datasets are still not selectable from the dropdown within Seqera Platform when using the dev branch. But I'm also not sure if the platform uses extensions for matching datasets, and instead opts for MIME-types. At least in the front-end, the extension is stripped from dataset names.

Yes, there are two issues here.

1. Lack of a pattern means Platform doesn't respond correctly (fixed in the linked PR)
2. Platform doesn't respond correctly to patterns with the multiple extensions required here. This is a bug in Platform itself and will need to be resolved there.

The only viable workaround right now is to reset the schema to use a single file extension in the pattern. You could override the schema yourself pending the fix to the above.