To my knowledge none of the dictionaries used by MeCab (IPA, Jumandic, or Unidic) includes romaji transcription of words. And actually there is no need for that:
There exist different transcription schemes (e.g. Hepburn, kunrei, 99 siki);
Information on the pronunciation of lexical units is already available (e.g. ドモ).
You have to write your own transcription routine... or look for an existing katakana-romaji transcription module (compatible with your transcription scheme)...