Saturday, October 01, 2016

Extending evolutionary theory? - Paul E. Griffiths

I will be attending the Royal Society Meeting on New trends in evolutionary biology: biological, philosophical and social science perspectives. I'll post each of the abstracts and ask for your help in deciding what question to pose to the speakers. Here's the abstract for Paul Griffiths's talk on Genetic, epigenetic and exogenetic information in development and evolution.

I outline an approach to measuring biological information where ‘information’ is understood in the sense found in Francis Crick’s foundational contributions to molecular biology. Genes contain information in this sense, but so do epigenetic factors, as many biologists have recognised. The term ‘epigenetic’ is ambiguous, and I introduce a distinction between epigenetic and exogenetic inheritance to clarify one aspect of this ambiguity. These three heredity systems play complementary roles in development and evolution.
Paul Griffiths is a philosopher of biology from Australia. He has been interested in genes for many years. I don't know what he's going to say about epigenetics. I suspect he can make a case for information in the sense that a naked DNA sequence differs in information from a methylated DNA sequence. (Restriction/modification is a good example.)

The problem with this kind of hair splitting is that ultimately the extra information (e.g. methylation) is provided by enzymes (methylases) encoded by genes in the genome. Epigenetics, per se, doesn't add any new information. It's just a consequence, or outcome, of the information already in the DNA.


  1. "Epigenetics, per se, doesn't add any new information. It's just a consequence, or outcome, of the information already in the DNA. "

    Yes, but whether it happens also depends on the environment. Which means that the sequence-plus-methylation-pattern reflects more than the information in the DNA.

    BTW: I'm a bit surprised. So far the people whose abstracts have been posted don't sound at all like Third Way types or people who demand a new synthesis, named after themselves.

  2. "Epigenetics, per se, doesn't add any new information. It's just a consequence, or outcome, of the information already in the DNA."

    I have to disagree slightly. The information on how to _make_ a methylase is in the DNA, but _where_ to put it seems to come at least partly from the enviroment.

  3. Here's a way of looking at it from the perspective of someone not so well versed in biology (although I grasp both the parity thesis and infotel).

    If the DNA is an information source in either Shannon's sense, or is a data structure more like in the sense of data sequences per Kolmogorov, or if it is regarded as a hybrid of the two as with Chaitin, or a semioitic structure per Sarkar: then in each case it is a structure (both sequential and conformational). The epigenetic factors can be separated into other sources or expressed in terms of data sequences (or even in semiotic terms as suggested by Sarkar).

    According to contemporary information theory, channels (signal pathways - if you like) can also be regarded as sources - and coupled with sources and taken as sources (a channel and a source together can be modeled a a source with its own different library).

    Now, it would seem that if one combines two separate sources, or a source and a channel, as a Shannon source, then one has a different information source and different information - necessarily. There are different sets of possible structures or configurations at the source, and thus a different library completely. The structure of the combined source is going to have a different library of symbols available, because there will be a completely different set of possible structures/configurations/conformations/sequences/symbols. Likewise with a data sequence considered as a structure in terms of a sequence of symbols per Kolmogorov (or even Turing), or as a total conformational structure: two separate data structures or sets of symbols are going to embody different information. Combining sequences or conformations has to deliver up new ones, and in the case of conformations and sequences both there is an argument available that these might be emergent and irreducible, at least in functional terms - if not otherwise.

    So on a quite big-dumb structure-based approach - it is hard to see how A. Epigenetic structures are not informational in their own right, and B. Epigenetic structures combined with DNA structures are not new sources with different information to either of their constituent sets of structures (DNA and epigenetic). Necessarily.