Gene interactions research studies have provided several useful
applications such as new drugs discovery that act on regulatory
pathways, and the development of tracking methods in the
dynamics of disease evolution within cells. GRNs estimation is
based on determining the genes that affect the expression of other
genes and on adequately describing these effects. Several methods
have been developed for inferring GRNs and producing hypotheses
about the presence or absence of interactions among genes;
hypotheses that can later be tested by laboratory experiments [15].
To address this issue, in addition to applications of genetic
algorithms [16], three main types of solutions have been proposed.
The first type limits the number of estimated genes. Even when
estimating a large-scale network, part of the network is often
attracted. The second type, a brute force approach, parallelizes the
estimation using high-performance computation. It is possible to
estimate large-scale networks when effective parallelizing is used
at the expense of costly computing equipment. The third type
improves the algorithm itself (e.g. greedy hill-climbing [GHC]),
avoiding the use of costly hardware but with the risk of being
trapped in local minima [6].