HUMAnN Collection Mapping Launch Failure

[Reki]

HUMAnN 3.9 runs successfully on individual datasets using:

• Bowtie2 unaligned read pairs (collection 646)
• MetaPhlAn predicted taxon relative abundances (collection 945)
• bypass taxonomic profiling

However, when selecting the full collections, clicking “Run Tool” returns immediately to the form and no jobs are created.

The same configuration works for individual samples.

Is there any restriction on collection mapping for HUMAnN 3.9 on Galaxy Microbiology?

Additional information:

The issue persists in a completely new history containing only:

- Bowtie2 unaligned read pairs collection

- MetaPhlAn predicted taxon relative abundances collection

The collections have matching element identifiers:

1_S1_001

2_S2_001

…

20_S20_001

When selecting single samples, HUMAnN launches and completes successfully.

When selecting the corresponding collections, clicking “Run Tool” changes the button to “Please wait…” and immediately returns to “Run Tool” without creating any jobs or displaying any error message.

Therefore the issue does not appear to be related to history size or collection naming.

[jennaj]

Hi @Reki

If you are working at UseGalaxy.eu, the odd behavior may have been related to this known issue, so please do try to run the tool again now. If resolved, please let us know!

Known issue

@Gabriel

Sorry for the inconvenient. We have done some changes that had some unexpected repercussions.

You should be able to run jobs now (:

---

Persistent issue

If you are working at a different server, or starting up the tool/workflow again now is still not possible, I’m wondering if there is a problem with how the collection data is organized, or maybe a problem within the workflow.

Some ideas about how to troubleshoot

1. A workflow needs a few changes to run on individual samples versus a collection folder, so there will be at least two distinct workflow versions most of the time.
2. The tool form has the different “select” options for individual, multiple, and collection input choices.
3. There can be some logic constraints for the different collection folder shapes too.

This topic has some screenshots and descriptions for how to check:

@jennaj

Data content versus shape

It is important to set the tool form options to reflect the content and shape of your data.

For the context of fastq reads from an RNA-seq sequencing project, this would include:

• Content: the type of sequence (rna-seq, dna, other) and the actual read itself (nucleotide bases, protein amino acids, other) then the quality score scaling scheme (fastqsanger in Galaxy refers to Sanger Phred +33 offset scaling).

  • Encoded in the assigned datatype format.
  • Compare to accepted formats on the tool form.

• Shape: how the data is arranged. Fastq data can have both ends of a read combined into a single file (interlaced paired end) or the forward and reverse reads can be in separate files (individual paired end).

  • Encoded in the collection folder’s element identifier levels.
  • Using a tool like Extract Element Identifiers is one way to review.
  • Tools under Collection Operations in the tool panel can be used to convert between shapes.

Quality assurance is performed to address the first. Sample organization is performed to address the second.

---

I’m sure you know all of this but it hasn’t been summarized in one place in a while, so I’m borrowing your question!

I’d like to review your use case! Would you like to share your history? → FAQ: Sharing your History.

This can be a subset of input data in a different history, or with tutorial data (good to compare against anyway). Just enough to trigger the behavior. Because the options will matter, please also post back some screenshots of the tool form with any relevant parameters set up. I’ll try to reproduce then make suggestions or we can probably come up with a workaround (and report the issues!). The tool version (top of tool form) should be captured, and the server URL will be in the share link.

You are correct that collections should be accepted but I’ll need to see the exact situation to offer advice. If you would rather post this in a direct message here instead of publically, please start one up! Then we can abstract the help back in the public threads.

Let’s start there!