They give confidence intervals in the text for about 51 MB(mega bases) which would be about 2 percent or so if that was across the whole mappable genome. Just a note that does not translate to 2% different genes or something, but rather that specific mutations they know can be traced back to a specific population. Most would have no effect.
From the article: "Compared with Australians and Papuans, the Negritos’ Denisovan ancestry was up to 46 per cent higher..."
46 percent higher than what? Do Australians have 0% of total? 50% of total?
> Denisovans apparently interbred with modern humans, with about 3–5% of the DNA of Melanesians and Aboriginal Australians and around 7–8% in Papuans deriving from Denisovans. [0]
The published academic article seems to estimate an upper bound of about 4%, but I might be interpreting that wrong. It is worth noting that the information is here based on single nucleotide variants which don’t tell us everything about the genome, but a convenient proxy for most of it.
It’s hard to sus an absolute percent but my best guess from some figures is 5% but I could be way way off.