used on plant genomes?

[colindaven]

Hi,

has this tool been used on plant genomes yet ? Would it be expected to work on modern crops (similar genomes and short divergence times) or is it more designed to estimate phylogenies across all grasses etc ? Ie., how sensitive is it and can it be tuned to this use case ?

Thanks

[ang037]

Hi,

Thanks for your interest in using our tool. We had tested it on a small dataset of 11 bamboo genomes, and it worked well. We have not explicitly benchmarked it on larger plant datasets, but it is expected to work on datasets with shorter divergence times (closely-related species), and ROADIES can be tuned for that setting.

The main parameters to adjust (in config/config.yaml) are the locus length (LENGTH) and LASTZ's alignment identity/coverage thresholds (COVERAGE, CONTINUITY, IDENTITY). For closely related genomes, I would generally use more loci and/or longer loci, and tune the alignment thresholds to recover informative regions while avoiding overly noisy matches.

The only case where I would be more cautious is extremely close strain-level comparisons or an intra-species dataset, where there may simply be too little phylogenetic signal for a species-tree method. But for broader grass/plant datasets, ROADIES should work reasonably well with appropriate parameters. Happy to help suggest starting parameters if you have a specific dataset in mind.