i'm afraid i'm going to have to disagree with ryan on this one.
firstly,
NEVER put overdrive, compression, or any other dynamic/distortion effects on the main output (i think he might have been misquoted there)...
Keep the output channel completely clear except for perhaps a frequency analyser (massively useful - i can't believe it isn't included in programs like reason... even more of a reason to step up to logic/cubase) and gain (this will enable you to bring the gain down a couple of db before bouncing out a pre-master - just so you can avoid any clipping that might occur). The pre-master should be as clean as possible for the purposes of mastering. The waveform will also give you a good impression of where the issues might be with your mix (look out for things like rogue peaks, squared-off peaks, too much of a dynamic difference between the drums and the 'music', etc).
secondly,
compression should not accentuate any peaks. the main point of compression is to bring the dynamics together so that peaks are reduced, while the quieter part will become louder (depending on where you set the threshold). Of course, there are different types of compression, and you should also look at experimenting with very fast attack/release times compared with slightly slower ones, to get the most out of a small amount of compression. It's easy to over-cook the compression so make sure you have solo'd the channel and listen carefully for the differences while watching the level meter.
as for transient shaper plug-ins, i've used the transient monster and flux bittersweet, and the latter seems to be a lot easier to control without over-doing it. The transient monster is fun for making extreme transient changes but the flux seems more detailed and professional.
if you are having trouble with peaking channels, you have exhausted all attempts at subtractive EQing, and you don't want to draw for compression or transient shapers (these will both 'colour' the sound - sometimes in an adverse way), you can always look at fading the audio in slightly (or if using a sampler, the actual sampler envelope itself). sometimes a fractionally longer attack time can not only make the peaks less problematic but can even make the samples sound clearer and warmer especially in very rigid quantised music. Another technique (esp with snares) is to play with off-setting and deliberately un-quantising layers so that the peaks of each layer don't all hit at once. This is especially handy with claps, and gives a more dilla-esque feel to the sound as well as giving you more volume to play with.
hope that's of some help
