In this work we present an integrated approach to the study of protein translation, based on Statistical Physics. We adopted three different but complementary perspectives: building hypothesis up from the data, modeling down from reasonable assumptions, and using computer simulations when everything else fails. In particular, we first analyze the mRNA sequences by means of inforation theory. We focus on the way the redundancy of the genetic code (the 61 sense triplets of nucleotides -the codons- encode for 20 amino acids) is utilized in the actual sequences, a phenomenon known as the codon bias.
We observe that it is not completely random, and encodes information in the frequencies and in the order of the codons.
With the scope of explaining these anomalies, we develop and analyze a family of stochastic models. Translation emerges as a systemic process, where the limited amount of resources in the cell couples the expression of the genes at a global level. We also suggest a game-theoretical interpretation of the codon bias. We finally attack the problem of cotranslational folding (i.e., the folding of the protein while it is still being translated). Specifically, we ask how this process depends on the order of the codons. This question is computationally very cumbersome. We thus propose a framework based on Markov chains, which allows the efficient simulation of arbitrarily complicate cotranslational folding mechanisms.
