It is obviously best to capture the source as best as possible but if you find you have too much fret/string noise and can't do another take then a de-esser can be used. However as they are not designed for this function it is not perfect. A de-esser reduces sibilant sounds (s, t etc), especially after a vocal has been compressed and so these, normally quiet unobtrusive sounds, are more noticeable. A de-esser is basically a compressor that is only working on a specific frequency, this works well for vocal sibilant sounds as they are normally at a very similar frequency throughout however normally each string/fret noise will be a slightly different frequency so you may have to set up multiple de-essers, it is a pain but you can get good results.