Issue: Inconsistent GFF Attribute Handling and ID Requirements

[elenitskorda]

Thank you for this amazing tool!

While preparing input GFF files for panHITE, I encountered inconsistencies in how the pipeline handles attributes and IDs. Some Prokka-generated GFFs without gene_id run successfully, while others fail unless normalized with AGAT. This raises questions about what attributes are truly required and how panHITE interprets them internally.

Observed Behavior:

• Two Prokka GFFs without gene_id processed successfully.
• Other Prokka GFFs required conversion with agat_convert_sp_gxf2gxf.pl to add gene features and gene_id.
• Documentation suggests gene_id must follow a prefix_suffix pattern (e.g., strain_AT1G01010), but the pipeline appears to fall back to other attributes in some cases.

Questions / Clarifications Needed:

1. What attributes are mandatory for panHITE?

• Is gene_id strictly required, or can ID/locus_tag substitute?
• Are gene features mandatory, or can panHITE infer from mRNA/CDS?

2. How does panHITE group homologous genes across strains?

• Does it rely exclusively on the suffix of gene_id?
• If gene_id is missing, what fallback logic is used?

3. Impact on results:

• If some files lack gene_id but run, does this affect pan-genome clustering or TE association accuracy?
• Could inconsistent naming inflate accessory genome counts?

4. Best practice for input normalization:

• Should all GFFs be standardized to include gene features and gene_id with prefix_suffix naming?
• Is there an official script or recommended workflow for this?