Necessary and sufficient conditions for a phenotype system to have a factor-union representation (Q793659)

The concept of factor-union representation of a phenotype system introduced by \textit{C. W. Cotterman} [in N. E. Morton (ed.), Computer applications in genetics. Univ. Hawaii Press (1969)] is defined as follows: Consider a phenotyp set \(\Phi\), i.e. a partition on the set of genotypes given at a locus, and a set X of ''factors''. Then a factor-union representation of \(\Phi\) is determined by an assignment of subsets of X to the alleles of \(\Phi\) such that (i) for any phenotype P in \(\Phi\) and any genotype ab in P \(f(P)=f(a)\cup f(b)\) (with \(f(P)=\cup_{x\quad in P}f(x))\) and (ii) f(P)\(\neq f(Q)\) if \(P\neq Q\). Condition (i) means that each phenotype is determined by the properties (factors) of the alleles of each of its genotypes. The author proposed an algorithm which allows to decide whether a given phenotype set has a factor-union representation or not, and if so, the corresponding representation is constructed by means of the theory of partially ordered sets. A further result is concerned with the size of the representation. The number of factors used usually corresponds to the number of phenotypes. Under certain assumptions, however, this number can be diminished considerably. The final section of the paper deals with multilocus systems and it is shown that finding a factor-union representation for multilocus systems reduces to the problem for one-locus systems.