I think you may be misunderstanding how the Bayh-Dole act works. The Stanford-Roche case looks like it hinged on a very particular set of circumstances involving a signle researcher working in both academia and industry, and the dispute was basically over who owned his work. The CRISPR case is different because the competing labs claim to have made the invention independently, with zero crossover. Even if Broad Institute were a for-profit corporation and not siphoning up vast amounts of NIH grants, the underlying dispute has nothing to do with Bayh-Dole or university IP policies, it's just a classic case of "who was first".