Hey check out an article on Gregor Mendel and how to do a punnett square. It will help you understand a lot about breeding for traits. The short answer is that you would have to cross a normal strain with your auto blueberry, then inbreed the f1 generation. Based on the punnett square, all plants in the f1 generation will carry the autoflowering gene, but won't express it because the normal flowering gene is dominant. Once you inbreed the f1 generation, 25% of the f2 generation will autoflower. Breed a male and a female that autoflower from the f2 generation and 100% of the f3 generation will autoflower.
This was how the first autoflowers were created. They mixed ruderalis (~1% THC) with a high potency strain (lets say 20% THC), and inbred it until 100% of the offspring autoflowered. However, being 50% ruderalis, and 50% high potency, they only had a moderate amount of THC. (lets say 10.5% based on the numbers above)
Then, start the whole process over. Use the f3 generation that was moderate THC (10.5%) and 100% autoflowering, and cross that with a highly potent normal strain (lets say 20% again). 3 generations of inbreeding later, 100% will autoflower, but this time they will have (10.5+20)/2=15.25% THC.
You can now understand why ruderalis has no effect (EDIT: hardly any effect) on potency after several crosses. One more cross with our hypothetical autoflowering strain up there and a highly potent normal strain could push the THC up to nearly 18%. This also shows why people are wrong when they automatically discount autoflowering strains as less potent.