running AHRD on chunked results (ie prior to concatenation) would probably be faster/less memory intensive

[adf-ncgr]

The older protocol used a built-in AHRD method (ahrd.controller.Batcher) for this purpose, but essentially all that does is take a templated version of the ahrd config yml and write one config per chunk and a shell script that invokes each batch sequentially. It would probably be easier to just do something similar but using nextflow-based logic instead.