Identification of domain pattern in human proteins using association rules

Abstract

In this paper, we introduce a formulated method to evaluate significance of each domain combination, which exploits association rules, and report an overview of domain combination by identifying domain pattern and analyzing their functional annotations. As proteins have evolved toward specific functions, domains, fundamental functional units, have high tendency to form patterns. Domain patterns must be significant domain combinations that have biological reasons to be assembled. Introduced method measures co-occurrence frequency and mutual dependency of domains in a domain combination, so it is useful to estimate whether a given domain combination is meaningful or not. Also we devised functional cohesiveness measure, which makes use of GO term annotation of domains, to investigate biological meaning of domain patterns. Based on the methods, we extracted domain patterns in human proteins and investigated functional annotations of them. From the results, we drew conclusionS that domains in human proteins form patterns whose members are highly affiliated to one another, and that extracted patterns tend to be associated with molecular function and biological process.