Implicit conversion in workflow causes results to differ if the input is compressed

[Delphine-L]

Describe the bug I am using a text transformation tool on a fasta input in a workflow. The text transformation requires a txt input. If the input is a fasta file, the output of the text transformation is a fasta, but if the input is a fasta.gz file, there is a conversion step that convert the fasta to tabular before the text transformation tool, and thee output of the workflow is then a tabular. The solution I am using to circumvent it is to explicitly asked the user if the input are compressed, and add optional decompression steps, but it adds complexity to the workflow and to the workflow form. This is related to issue #18709, but I open a new issue because one of a solution discussed was suggesting both : fileA (as fasta) and fileA(as tabular) for the inputs, but it wouldn't solve the issue inside the workflow.

Galaxy Version and/or server at which you observed the bug Galaxy Version: Main 24.1.3.dev0

Browser and Operating System Operating System: macOS Browser: Chrome

To Reproduce Steps to reproduce the behavior:

1. Import workflow https://usegalaxy.org/u/delphinel/w/test-implicit-conversion
2. Import history https://usegalaxy.org/u/delphinel/h/test-implicit-conversion-fastagz-to-txt
3. Run the workflow on both the compressed and uncompressed fastas (datasets 1 and 2)
4. Observe that the results are different (datasets 3 and 4)

Expected behavior Suggestions of solutions :

• Add additional constraints for tools inside a workflow (if input is txt, I could be able to restrict it to fasta)
• Being able to test the format of an input, in that case the optional decompression step could be triggered by the detection of a compressed fasta
• Being able to constrain a workflow input as compressed or uncompressed

[Delphine-L]

Strangely the issue doesn't happen on vgp.usegalaxy.org, the fasta.gz is converted to fasta before being used as an input:

[natefoo]

A bit more detail from my testing:

• Run as a tool outside of a workflow, the tool's input field shows 1: Compressed fasta (as tabular) and the conversion is always fasta.gz -> fasta -> tabular, and the tool outputs tabular.
• I can reproduce @Delphine-L's findings when run in the example workflow:
  • On usegalaxy.org, the workflow input form shows 1: Compressed Fasta (as fasta) but it converts fasta.gz -> fasta -> tabular and outputs tabular, the same as when running as a plain tool.
  • On vgp.usegalaxy.org, the workflow input form also shows 1: Compressed Fasta (as fasta) but it converts fasta.gz -> fasta -> and outputs fasta.

usegalaxy.org and vgp.usegalaxy.org do run slightly different configs but the same copy of Galaxy itself, the same datatypes_conf.xml (the sample shipped with Galaxy at the revision it is running), and the same database, and I don't see any differing config options that would affect this.

[mvdbeek]

I'm gonna close this as a duplicate of https://github.com/galaxyproject/galaxy/issues/18709 and prioritze a fix for that.