FUNGuild: Best practice for OTUs with multiple guilds (no primary “|”), and handling “Plant Pathogen” priority?

Other Bioinformatics Tools
1

Hi all,

I’m working with fungal OTUs/ASVs (ITS) from a QIIME 2 workflow and assigning functional guilds with FUNGuild. I’ve been stuck on how to handle multiple guild assignments, and I’d really appreciate guidance or references to common practice.

What I think I understand

  • When a vertical bar (|) is present, the guild inside the bar is the primary assignment, and it’s reasonable to retain that assignment when the confidence is “Probable” or “Highly Probable.”
    Example: Endophyte-Plant Saprotroph-|Undefined Saprotroph| → keep Undefined Saprotroph if confidence ≥ Probable.

My questions

  1. Multiple guilds without a primary bar
    Example: Endophyte-Litter Saprotroph-Soil Saprotroph-Undefined Saprotroph (no |).

    • Should these be treated as ambiguous and excluded from guild-level summaries?

    • Or should I retain them in some principled way (e.g., keep only the first listed, split counts fractionally across listed guilds, or duplicate the OTU across categories)?

    • Is there a recommended, citable best practice?

  2. Prioritizing specific trophic modes
    a) Plant Pathogen priority: If “Plant Pathogen” appears among multiple assigned guilds (but not marked as primary), is there any accepted rule to prioritize it as primary (e.g., for risk-relevant analyses)? Or is it better to flag it separately while not overriding the absence of a primary assignment?
    b) Saprotroph vs Symbiotroph (sometimes typed “symbiotroph”): If an OTU is co-assigned to Saprotroph and Symbiotroph, is there a community norm to prioritize one over the other (e.g., always prefer Saprotroph), or should such OTUs be retained as mixed categories (e.g., “Saprotroph–Symbiotroph”), flagged, or excluded from single-guild summaries?

  3. Dropping highly ambiguous OTUs
    Would the following rule be considered reasonable by the community and reviewers:

    “All OTUs that had three or more possible guilds were classified as undetermined and dropped from guild-specific analyses.”
    If this is acceptable, are there references or prior forum threads I can cite?

Additional context

  • I already filter to retain only guild assignments with confidence = Probable/Highly Probable.

  • My goal is to produce transparent, reproducible guild-level summaries without over-interpreting uncertain assignments.

Thanks in advance for any advice, example workflows, or papers that outline community norms for these cases!

3

Looking at this post and wondering if you found an answer! having the same issue