files_refmust have the same TE as your metabolite data (for eddy-current correction). files_w can have shorter TE, which means there is less T2 weighting on the water signal.
If you don’t have files_w, Osprey will still provide relaxation-corrected, water-scaled metabolite level estimates (based on files_ref).
I’ve never seen real-life ERETIC data, but I’d be curious to integrate the option to process it into Osprey. If you have data and analysis code that you and Laurent would be willing to share, that would be a great basis to start from.