assembly: purge_dups removes too much sequence

Question

[colindaven]

I have tried out purge_dups and found results to be puzzling.

First, there seems to be a silent bug where contig fasta headers are ignored (symptom - bed files are empty).

Solution - use simple contig fasta headers, like sample_contig1 and not sample-complex-name_x1-_contig-1.

I analyzed plant genomes, which are repetitive, and found >80% of the contigs would be removed (700 mb reduced to 60 mb). This is surely too much.

I am using ONT and not Pacbio HiFi for my assemblies, so this could be one problem.

Has anyone optimized purge_dups for either plant genomes or nanopore ? It has over 850 citations and has been used on plant genomes widely before, yet there is no recommended parameter set for plants.

Thanks