Wide range of the genetic sequences of homo sapiens has led to the formation of the new field of bioinformatics which is the integration of molecular biology, computer science, and mathematics. New developments in bioinformatics aim to detect genetic patterns of complex diseases which were previously believed incurable. A major part of these developments is owed to gene mapping which deals with determining disease susceptible regions. Most of the existing work is statistical and based on assumptions. However, since the genetic information is stored in a digital manner, gene mapping has close relationship with information theory and this relation is the base for more recent methodologies. One of the disadvantages of current approaches is their futility against epistasy phenomenon amongst genes. In this work, algorithms are proposed that are effective in handling this phenomenon. These algorithms which also include relevance-chains algorithms are compared to the existing ones. Furthermore, a new criterion or principle is proposed for association mapping which is the generalized form of gene mapping and is discussed from different aspects. Results from the relevance-chains algorithms on real data show that AMD disease has many epistatic models amongst its markers whereas Parkinson’s disease does not follow such models.
