Can Pandagma be used for pan-genome analysis of a specific plant gene family?

[changchuanjun]

Dear authors,

Thank you for developing Pandagma—a fascinating and valuable software! I have a question regarding its functionality: can Pandagma be used for pan-genome analysis of a specific gene family?

To elaborate, I aim to conduct a pan-genome analysis of the WOX gene family across 60 Oryza accessions, with the goal of identifying core, softcore, dispensable, and private WOX genes. My plan is as follows: first, I will identify all WOX gene family members in each of the 60 accessions, resulting in 60 FASTA files (each containing the full set of WOX protein sequences for one accession).

I wonder if Pandagma can process these 60 WOX protein FASTA files as input and generate an output similar to OrthoFinder2’s "Orthogroups.GeneCount" (formatted as described below). This output would then allow me to classify the WOX genes into core, softcore, dispensable, and private categories, as I currently do using OrthoFinder2.

Looking forward to your reply !

Orthogroups.GeneCount content example:

[StevenCannon-USDA]

Hi @changchuanjun,

That is an interesting question. I have never thought of using pandagma in this way. I expect it would work, if you wanted to make a set of files with WOX-like genes for each accession. For each accession, you would need three files: one with the primary CDS sequences, one with the primary protein sequences, and one with the gene locations (as a bed file). You would also need a config file.

However, personally I don't think I would do it this way. The selective approach seems backwards to me, and bypasses some advantages of the main programs. Namely: both the pangene and the gene family workflows allow competitive placement of genes into either pangenes or into families (depending on the workflow).

I would instead do a run of pandagma pan on the full set of annotation files (CDS, proteins, and coords/annotations). Then do a lenient BLAST search against the consensus sequences to find the WOX-like pangenes.

Or if you wanted to find the WOX-like gene families across a broader taxonomic scope (including other monocots besides Oryza), then do pandagma fam with a set of carefully selected rice representatives and non-Oryza outgroup species.

One comment and recommendation: 60 accessions is a lot. The initial comparison and construction steps (for both the pan and fam workflows) are expensive, because they do an all-by-all comparison. That would be (60x61)/2 = 1,830 comparisons. If you can identify a subset of the 60 that represent most of the diversity that you want to capture, the job will run a lot faster, and the quality may be greater than if all annotations are placed into the "main" set. You would want to include the highest-quality annotations, and the ones that are likely to be distinct. The remaining ones could be included in the annotation_files_extra_constr and/or annotation_files_extra_free sections of the config. See the section of the README about preparation of the config file -- step 3 under https://github.com/legumeinfo/pandagma/?tab=readme-ov-file#example-run-for-the-pangene-workflow-

[changchuanjun]

@StevenCannon-USDA Thank you for your detailed and helpful answer. I understand. Thank you again.