Is it possible to understand what a gene/junk dna does by observation? If so, perhaps it can be rewritten from scratch without understanding its implementation.
Understanding implementation that is a product of evolution is extremely hard. It's a real mess.
Someday maybe. There are something on the order of 25,000 genes. Lots, maybe even most, have variants, which increase the chance of disease, particularly with certain variants of other genes. Many genes are triggered only at certain times in the embryo's development and in an adult may have an entirely different purpose. So here you have a massive super computer with 25,000 objects and variants whose function varies over time. I don't think we are even close to figuring it out. For example, simulating the folding of a single protein takes a super computer.
Understanding implementation that is a product of evolution is extremely hard. It's a real mess.