Imperial Italian samples from Rome can be modeled as descending overwhelmingly from the mixed Greek-Anatolian population of Hellenistic Anatolia. I've only included non-outlier samples, representative of the average population.

The results were pretty much identical whichever post-Bronze Age Anatolian source I used, but the Hellenistic era source gave the best fit and is the most historically plausible, since the Romans conquered Hellenistic Anatolia.

I have no idea what the authors of the old Roman & Etruscan genetic studies were thinking when they modeled Imperial Italians as half Roman Republic + half Levantine.

It looks like they have no Republic-era central Italian ancestry at all - although G25 often confuses Roman Republic ancestry with Balkan or Continental Celtic ancestry, due to their genetic similarity. Any "Levantine" ancestry could also be Eastern Anatolian. It's possible that my model is terribly wrong somehow, but I doubt it.