Information theory and error-correcting codes in genetics and biological evolution
pp. 299-345
in: Marcello Barbieri (ed), Introduction to biosemiotics, Berlin, Springer, 2007Abstract
As semiotics itself, biosemiotics is concerned with semantics. On the other hand, the scientific study of communication engineering led to the development of information theory, which ignores semantics. For this reason, many biologists thought that it would be useless in their disciplines. It turns out however that problems of communication engineering are met in biology and thus can only properly be dealt with using information theory. As an important example, the faithful transmission of genetic information through the ages is a difficult problem which has been overlooked by biologists. Cumulated errors in the DNA molecule due to radiations and even to its own indeterminism as a quantum object actually perturb its communication through time. A simple information-theoretic computation shows that, contrary to the current belief, the genomic memory is ephemeral at the time scale of geology. The conventional template-replication paradigm is thus not tenable. According to a fundamental theorem of information theory, error-correcting codes can perform almost errorless communication provided certain conditions are met. Faithful conservation of genomes can thus be ensured only if they involve error-correcting codes. Then the genomes can be recovered with an arbitrarily small probability of error, provided the interval between successive generations is as short (at the time scale of geology) as to almost always avoid that the number of cumulated errors exceeds the correcting ability of the codeThis paper presents an intuitive outline of information theory and error-correcting codes, and briefly reviews the consequences of their application to the problem of genome conservation. It discusses the possible architecture of genomic error-correcting codes, proposing a layered structure referred to as "nested codes' which unequally protects information: the older and more fundamental it is, the better it is protected. As regards the component codes of this system, we notice that the error-correcting ability of codes relies on the existence of constraints which tie together the successive symbols of a sequence. It is convenient in engineering to use mathematical constraints implemented by physical means for performing error correction. Nature is assumed to use to this end 'soft codes' with physico-chemical constraints, in addition to linguistic constraints that the genomes need for directing the construction and maintenance of phenotypes. The hypotheses that genomic error-correction means exist and take the form of nested codes then suffice to deduce many features of the living world and of its evolution. Some of these features are recognized biological facts, and others answer debated questions. Most of them have no satisfactory explanation in current biology. The theoretical impossibility of genome conservation without error-correcting means makes these consequences as necessary as the hypotheses themselves. The direct identification of natural error-correcting means is still lacking, but one cannot expect it to be performed without the active involvement of practising geneticists. The paper also briefly questions the epistemological status of the engineering concept of information and its possible relation to semantics. Roughly stated, information appears as a necessary container for semantics, providing a bridge between the concrete and the abstract