This guy mathed out the optimal ratio as 3:1.
While I like the math, there are some additional considerations worth mentioning.
1. Space devoted to fuel can be handled by fuel tankers (with some cost to security and convenience). Space devoted to engines is less easy to offload (although carriers and tugs both work). This is the major reason why my own designs tend to carry less fuel than the above math suggests.
2. Engines of less than 100% fuel consumption are disproportionately less costly to research and field. This can only occasionally be taken advantage of with military designs, but is extremely important for civilian vessels.
3. Researching engines, especially big ones, can get costly. There is a significant price tag attached to per-design optimization. I tend to optimize for a fleet, using 2-3 types of engines, rather for each of perhaps 6-10 military ship classes.