Department of Biophysics, Faculty of Science, Kyoto University, Oiwake-cho, Kitashirakawa, Sakyo-ku, Kyoto 606, Japan
(Received April 28, 1993; Accepted June 25, 1993)
Keywords: Development, Axis Formation, Drosophila Melanogaster, Neural Network
Abstract. To understand a phenomenon in developmental biology, it is essential to grasp not only the function of each factor but also the behavior of the whole system. I present a method to investigate the interaction between each factor related to some phenomenon and predict unknown parts of it from experimental data. First, a feedforward neural network is used to embody the input-output relation of the system and the effect of each factor on the output is analyzed. Second, an orderness analysis that deduces the order of working of the factors in the system is presented. I applied these methods to early embryogenesis of fruit fly, Drosophila melanogaster. The causes and effects of the genes involved in the dorso-ventral axis formation are clearly shown.