What I don't know about genetics and gene therapy

Once, several years ago, I did an online university course on genetics, just for the hell of it. At the time, it all seemed to make sense, and I did fine on the quizzes and tests. Recently, though, the more I think about it, the more I realize that there are some pretty fundamental parts of the workings of genes that I just don't understand. On the basis that a little knowledge is probably a dangerous thing, I'm here to appeal to the Internet community to set me straight.
Here's what I think I understand:

• The nuclei of pretty much every cell in our bodies each contain 46 chromosomes, which are made up of long strands of DNA, cunningly twisted, coiled and folded into a very small space. Without going into too much detail, this DNA is in turn composed of billions of nucleotides, which are linked pairs of the four chemical bases (A, G, C and T) supported by structural sugar and phosphate molecules.
• Shorter sections of these long strands of DNA, from a few hundred to a couple of million base pairs each, arranged in a specific order, make up individual genes, of which we have around 20,000 to 25,000. We have two sets of genes, one inherited from each parent, and it is our genes that provide the instructions for building the 10,000 or so different complex molecules called proteins that make up and regulate the structure and function of the body's tissues and organs. Over 99% of these genes are identical in everyone, and these are what make us human (rather than chimps or beluga whales). Less than 1% of our genes are different between individuals, the variations in genes being called alleles, which is what makes one individual different from another.
• Each individual cell, though, only "turns on" (or expresses) a fraction of the genes in its DNA, the rest remaining "turned off" (or repressed), and this gene regulation is what causes a particular cell to function as a brain cell or a muscle cell or a heart cell. Just to confuse things, the expression of genes can also be affected by environmental influences like diet, pollution, etc, by a process known as epigenetics, and genes can also mutate over time if they copy incorrectly, or if an mutated gene is inherited at birth, but let's not go there for now.

So far so good. Where I started questioning my knowledge of genetics, though, was when I started reading about gene therapy, initially with reference to my wife's Parkinson's disease, and I could not understand how a knowledge of genetics could be used to make a concrete effect on an individual's body. This is all part of the wider question that I have about how all the millions of dollars being spent on the genetics of various diseases is actually going to result in practical steps in treating those diseases.
So what do I know about gene therapy? As the name suggests, it refers to techniques, still largely experimental, that use genes to treat or prevent disease. This may be by replacing a mutated gene that causes disease with a healthy copy of the gene, getting rid of or making inactive a mutated gene that is not functioning properly, or even introducing a completely new gene into the body in order to fight a particular disease. Apparently, inserting a new gene directly into a cell usually does not work, and so a carrier called a vector, such as a modified virus, can be used to deliver the gene (yes, just like on Helix!).
Right. But, doesn't this mean that the new gene would need to be inserted into every cell in the body (there doesn't seem much point in just one or two cells having an altered copy of a particular gene)? Is it even allowed for different cells to have a different set of genes (I always thought that every cell had exactly the same set of genes/DNA, just that some cells expressed some genes and some others)? Or is it, for example, sufficient for all the lung cells to host the altered genes in order to have a beneficial effect on a lung disease (not that that would be particularly easy in itself)?
In fact, all of this begs an even more fundamental question: how does a particular cell know which genes to turn on, and which not, in the first place? How does one hair follicle on a black and white cat know to switch on the white hair gene and the one next door to it to switch on the black hair gene? And, for that matter, could a cell in your elbow have a gene for blue eyes switched on, and, if it did, what difference would this make? The more I think about it, the more questions I could come up with.
Now, I know that not many people ever read this blog, and most of those are probably not well-versed in genomics. But, nothing ventured, nothing gained: maybe some kind soul will put me out of my misery.

UPDATE
My daughter actually came through for me on this one (she's doing a Biology major, and is currently interning in a genetics lab). Perhaps predictably, I got a lot more information than I either need or understand, but the quick summary is that, as I suspected, gene therapy is still pretty much pie in the sky, a cool idea whose time has not yet come.
Viruses can indeed be used to introduce new genes into cells, but the science is as yet very inexact. For example, a gene might end up integrating into a part of the DNA where it can't be expressed, or where it is inappropriately expressed all the time, or even into the middle of another gene, thus screwing up both genes. There are other methods of injecting DNA into cells, but they just introduce the new material into the cytoplasm of the cell, not the nucleus, and so it will gradually degrade over time. Yet another line of approach might be to introduce new genes into pluripotent stem cells, and then engineer these into the right kind of cell for the required purpose (all of which can indeed be done), but then the problem of delivery still remains.
And, yes, the new genes do have to be introduced to the right part of the body (whether that be bone marrow, muscles, lungs, etc) to have the desired effect, otherwise they will not be able to be properly expressed and will not be able to carry out the required function, and at present there seems to be no clever way of doing that. As for how to deliver new genes into the central nervous system or across the blood/brain barrier (for neurodegenerative diseases, for example), that is a whole new set of problems. And, even once introduced, it is still not clear whether the new altered cells will persist and proliferate, or just degrade and fade.
Finally, on my more general questions at the end:
The way in which different genes are expressed in different parts of the body is, well, complicated, involving tissue-specific transcriptional regulators, factors and enhancers, etc. I don't think I even want to go there. Because of these regulators, though, it is apparently very unlikely that a gene for blue eyes would find itself turned on in an elbow, to use my random example. And even if it did, and a blue pigment was created in an elbow cell (whether this be in bone, blood, cartilage or whatever), it would have no way of physically expressing itself, and so would just degrade and be removed.
The colour of cat hair is apparently a whole subject unto itself, and incredibly complicated and convoluted. Those interested would be better served by going to the detailed Wikipedia article on the subject, than having me try to explain it.
So, it seems that sometimes a little knowledge may actually be safer than getting totally tied up in a whole bunch of things I really don't understand, and don't have several years to devote to figuring out. Sad, but true.