The noise in gene regulatory networks is analyzed. First, biological background of the gene
expression is presented. Next, models of a single gene expression and autoregulatory protein
are introduced. Main result of the thesis is the study of noise in a two-gene regulatory
network. Analytical formulas for noise in five network architectures are found. Concept
of switching noise is introduced to measure an operational level of noise and to compare
noisiness of network architectures. More and less noisy connectivities are found and the
energy cost of regulation is estimated. Subsequently, two negatives feedbacks are introduced
into a network. Main results of the thesis show: (1) which onnectivities are the least noisy
ones, (2) that decrease in noise strength requires a large amount of regulatory molecules
(which in the process of synthesis involve large quantity of energy), (3) that autoregulation
is an economical method of noise reduction in a two-gene network.