Comparing these two on their careers, when you just admitted that a back with a higher proportion of carries would be less efficient on this metric, is very contradictory.
Simple thought experiment: What would D Williams points per touch look like if his proportion of rushes/receptions was roughly the same as Saquon's?
Saquon's rushes/reception = 261/91 = 2.87. You would have to add ~126 carries to D Williams' career totals to get to approximately this proportion (183 carries + 126 carries/108 receptions) = 2.86. Using his career averages, and with generous rounding on TDs, add 504 rushing yards and 5 TDs to his career totals. That's about an extra 80 fantasy points.
Take his new touch total/new fantasy points = (291+126)/(360+80) = 0.95 points per touch. Significantly different from above, clearly not on Saquon's level, but not too bad, either.
Now I don't think this quick and dirty thing above is the right way to think about this, but I just wanted to show that the metric you are using is flawed and will always over-inflate pass-catching backs with limited rushing involvement. Here's another example - your metric would say James White is the best RB in the league. Do you really believe that?
Here's a suggestion: Either look at points per rush and points per catch separately (safer) or come up with a more robust way of combining the two together with some sort of weighted average or scaling to limit how much your metric overvalues pass-catching backs (will always be a little flawed, but would at least be a lot better than what you are using now).