This study is to investigate structural and functional features of English lexical bundles in the Biomed Corpus as a Life Science Corpus through key clusters analyses. Previous studies have identified lexical bundles using their different criteria, but in this study lexical bundles were extracted and identified in a limited number of frequent phrases in comparison to a reference corpus using key clusters analyses. Structural and functional classification of lexical bundles was made following recent work by corpus linguists such as Biber and Barbieri. We found that the lexical bundles that we dealt with represent both positive and negative keyness in value in the list of key clusters, just like keywords analyses. The article examines and discusses the distribution of the frequence category of lexical bundles, the ratio of type and token of nouns in the structures of ``the+Noun+of the`` and ``in the+Noun+of``, various types of passive constructions, and epistemic stance expressions standing out in functional distribution of lexical bundles in comparison to a reference corpus.