Personally, I would recommend doing an original universe TC, purely because if it starts to get overwhelming, you can cut stuff, polish what you have, and release if you feel you won't finish. The three most recently released TCs, Colosseum, Acheron (previously Dark Star), and Ashen Galaxy, ALL did that. Doing a pre-existing universe, while useful because you don't need to invent the lore on your own, among other things, means you have to match their size, which could be tiny or HUGE. And if you burn out before reaching that size, well, you're stuck with a clearly unfinished product, no matter how you polish it.
That said, whether you continue to pursue BSG or not, there's several ways to go about doing the TC. I'll list the extremes, most people do a mixture of these.
1. Make all of x, then all of y, etc. - Seems to be the most common approach and usually starts by placing all the systems. I personally dislike this method for various reasons. One, once you have your huge set of empty star systems down, it'll look incredibly overwhelming and discouraging to actually fill them. Two, you may not even need all of the Xs you created. Three, you'll inevitably do something later that'll make you retweak things, which can lead to a ton of work, as much as redoing those 300 whatevers you made earlier. Four, ships are balanced by their own stats, their weapons, AND their outfits. Make all of any of those before the others and, well, see #3. And finally, five, if you need to downsize to complete it, you'll likely have to cut out more things here than in other methods. And trust me, cutting stuff is never easy, you will not want to remove things if you can help it.
However, this method does have it's advantages. If you're working with a small galaxy (which many people aren't), it's a good way to divide the tasks. It also works well for team assignments "you do all of x, you do all of y, and you do all of z". And, finally, for those people who get sidetracked always wanting to add new things, you can put all of your stuff down and say "Ok, I'm not adding any more Xs, no matter what!"
2. Start from the core - Start by making everything essential. Does Earth feature prominently in the plot? It gets added early. What about Mars? It's just fluff, we'll add it later. Does the player start in a Shuttle? That's the first ship we're making! Does it come with a weapon? That gets added too! And so forth. I like this method because, if worse comes to worse, you got a good chunk of the stuff you need finished to be completed (and, consequently, you'll probably have a "playable" plug early, allowing you to test things as soon as you make them). If you need to cut, you won't have much to remove and for the rest of the extras that would have had to go, you never got around to adding them, so no loss there!
It does have some issues, namely with big and team projects. For the former, a big TC will take awhile to start taking shape since the core to it's gameplay and stories will take longer to make. With a team, you'll have to define what the core is, which isn't easy unless you spend a good deal of time creating planning materials, documents, etc, and with such a spartan allowance of things to work on, people may step on each other's toes (you might want to allow some to do fluff work in those cases). And third, if you have to cut and release early to avoid going the way of almost every other TC, your released product might seem a bit empty and lacking. And finally, this can lead to some very convoluted resource ordering that, depending on how hard it makes things, may force you to do a lot of work later reorganizing your resources into more sensical ID groups.
3. One sector at a time - Let's assume you were making EVN. In this case, the Federation is a sector, the Aurorans are another, and so forth for all the other govts. Yes, even the small ones with one system. In this approach, you'd, say, start work on the Federation and create EVERYTHING needed to have a fully functional Federation, sans missions that require the other sectors. It's sort of a "putting together a puzzle" approach to TC making, except you have to build each piece of the puzzle before you can add it. It's helpful because you'll see definite progress as your completed government fleets start cruising the spacelanes overhead as you work on another one of their planets. And if you're not going to have time to finish all the pieces to your puzzle, then you have several completed ones you can put together to make a smaller picture.
It does have problems though. First, depending on how you are, you may be tempted to polish the current sector you're working on so much you never get started on the others. Second is the whole "gotta start over again" feeling when moving to the next sector, which can be discouraging for some. Third is when you cut your losses and downsize, you may still have a good chunk of work to do to make the pieces you have not feel like an incomplete puzzle, such as revision, retconning, and adding in those storylines you hadn't gotten to yet, except altered to not include whatever you cut. And the fourth can be varying levels of quality for each sector. If you come up with a cool, new idea that makes sector #4 awesome, you'd either have to work it in to the other sectors to make them awesome or have them risk feeling like older, less polished material. On the other hand, if you're slowly burning out on later material, it may seem rushed or less polished, resulting in a EVN like situation. Have you ever noticed that just about everything happens in Federation space? Almost every storyline, minor and major, starts there, leaving almost no reason to wander about in Auroran space. Polaris space is worse since there isn't even pirates to hunt, although that's a lore reason.
Not to accuse ATMOS of burning out or anything, as that may be an intentional design decision, albeit one I'd disagree with if I had been working on Nova.
4. Storyline first - Start out with the main storyline(s) and create everything needed for them. Mainly an alternative to #2 and has similar advantages and disadvantages, albeit for different reasons. If the main draw to your TC is the plot, then this is identical to #2 anyway.
5. Other approaches - I'm sure there's some I'm forgetting or not aware of.
