You're on the right track, but you're correct. Building a MYIt's going to make a new dataset output. You can then take that and plug it into any relevant workflow.

If you're looking to remove specific entries from a historic append chain, there's another option. There's a hidden function on DOMO's dataset. If you add ?_f=dataRepair to the end of a dataset url, a new tab called Data Repair will pop up. In there, you can see all the uploaded data versions.