It appears that libshuff will not work with column= name= as distance matrix input. Is this true? Is there a workaround? Is there a reason?
No, it’s not possible and we are unlikely to make this compatible. A couple of reasons… (i) s-libshuff is a weird method (see my ismej paper), (ii) with the number of sequences that you probably have you will find significant differences, (iii) it would probably be computationally prohibitive to run the method with large datasets.
Hope this helps,