Largest dimension of the room can determine the accurate reproduction of the lowest frequency of your audio system. At the sea-level where speed of sound is 1,130 ft/sec, longest dimension of the room should be about 56.5/2 = 28 ft to capture one half of the 20 Hz audio signal. This calculation is based on formula v = fxw where v = velocity of sound, f = frequency, and w = wavelength. 18 ft long room can reproduce 30 Hz and 14 ft one can reproduce 40 Hz accurately. Your speaker specifications (3dB cutoff) can be used for this calculation. This is not a requirement for subwoofers as they reproduce lower frequencies by bouncing the sound off walls many times. Above theory applies only if you want accurate reproduction of the sound.
Show-boxed shaped rooms are subjected to many room modes where you will hear certain frequencies amplified at certain locations and dead spots on other locations. A trapezoidal shaped room can minimize most of these room modes. If you have room where roof/ceiling is at an angle, then it will most likely be a trapezoidal shape or one closer to it. I understand this may not be an option for some, but at least in theory those room shapes are better than shoe-boxed shape ones.
This is just a start. Next you have room treatment which is a very big topic covered very well on this site. Do a search.