Learning processes in hierarchical pairs regulate entire gene expression in cells.

Expression of numerous genes is precisely controlled in a cell in various contexts. While genetic and epigenetic mechanisms contribute to this regulation, how each mechanism cooperates to ensure the proper expression patterns of the whole gene remains unclear. Here, I theoretically show that the repetition of simple biological processes makes cells functional with the appropriate expression patterns of all genes if the inappropriateness of current expression ratios is roughly fed back to the epigenetic states. A learning pair model is developed, in which two factors autonomously approach the target ratio by repeating two stochastic processes; competitive amplification with a small addition term and decay depending on the difference between the current and target ratios. Furthermore, thousands of factors are self-regulated in a hierarchical-pair architecture, in which the activation degrees competitively amplify, while transducing the activation signal, and decay at four different probabilities. Changes in whole-gene expression during human early embryogenesis and hematopoiesis are reproduced in simulation using this epigenetic learning process in a single genetically-determined hierarchical-pair architecture of gene regulatory cascades. On the background of this learning process, I propose the law of biological inertia, which means that a living cell basically maintains the expression pattern while renewing its contents.