I've tried a few different things but not every permutation by any means. It is instructive to look at a few different sprockets that are meant for use in derailleur systems;
SRAM sprocket for brompton 6s setupThis looks like a standard IGH sprocket that someone has attacked with a grinder. Like UG sprockets it has some (but not all) teeth with truncated tops. It is directional, but only just, because of a few teeth with an extra cutaway. It kind of gives you hope that you could attack an IGH sprocket with a grinder an make it work in a derailleur system. Since there are only a few teeth which are cutaway, it may work in reverse of the marked direction; I have not tried this.
current Brompton 6s sprocket This looks nigh-on identical to many HG type sprockets and uses an HG freehub spline. It is also directional because the teeth are cutaway (not just because of the spline).
One thing that may catch you out is that chains will happily run on sprockets (and chainrings) that are a fairly snug fit in the chain, but the chain won't easily shift
off them unless there is enough clearance for the links to twist slightly.
So for SS use, a nominally 3/32" ( 2.38mm) sprocket can be used (which is usually ~2.2 to 2.3mm in singlespeed form), but this isn't going to work well for a derailleur setup; the teeth (at least) need to be thinner; in the range 1.9 to 2.0mm is normal for an 8s chain to work.
DX sprocketSo if you want a SS type sprocket (like the DX one) to work with an 8s chain as part of a derailleur system it will do provided you are prepared to do some work with a grinder as necessary. For example you can make the teeth truncated in some cases, and slimmer too, as necessary. You can also grind them (with two angled facets near the top) to give the functional equivalent of UG-style 'twist teeth' .
BITD I ran out of the Sun Tour sprockets with the correct tooth chamfers to work in some positions on 7s freewheels, but I did have ones with the same centre fitting. I experimented with grinding the tooth tops to give a 'twist tooth' type profile and this gave much improved shifting, regardless of where they were situated. They also wore faster near the tooth top (above the most loaded part of the tooth), which (similarly to HG sprockets I suppose, but for slightly different reasons) meant that they were less prone to becoming obviously hooked. They were still 'ramped' though, and at the time I didn't understand why the chain skipped....
cheers