Is it possible to identify subsets of bands which are common to groups of isolates, which may then be used in turn as bands for identification of species, serogroups etc.?
Yes, starting from a composite data set, you can perform a Band matching and subsequent polymorphism analysis. You can show the polymorphic bands only (Bandmatching > Polymorphic classes only) and so the band classes that are common to all patterns are shown.
An even more refined solution is to do a discriminative search. Within the composite data set you can make a selection of a group of organisms and then Edit > Bring selected entries to top. Then select Composite > Discriminative characters. The characters are reorganized in such a way that those characters positive for the selected entries and negative for the other entries occur left, while those characters negative for the selected entries and positive for the other entries occur right.