The example of single-sided amplitude spectrum at
doc fft
shows clearly how to get the associated frequency axis on a plot given the known sampling frequency and sample time (number of samples). As noted there, the fft as returned by fft is double-sided (includes both positive and negative frequencies), hence the NFFT/2+1 factor for single-sided only.
As for the second question, sure, any change in the frequency content will show up when the transform is inverted back to time domain; by theory that will be the convolution of one in the other domain is the product in the other. You'll have to know what the desired effect is a priori in order to produce a desired result, of course.
