Two Telling Tales

I am writing this despite a reluctance to add to the chorus of criticisms of the scientific publication process — an industry full of easy over-generalisations and proposals to discard babies with their bathwater. But a couple of recent publications do, I think, illustrate some recurring problems in the scientific “literature”. One of these is highlighted by the Graur et al. [1] paper in Genome Biol. Evol. which reports, in a gleefully witty style, a needed critical analysis of some of the much-publicised claims of last year’s Nature ENCODE paper [2]. The Nature paper famously made a case, based on the ENCODE consortium’s analysis of the human genome, that most of genome sequence is, in some sense, “functional” — an assertion that was very widely picked up and promoted in the general media, which claimed that the paper sounded the death-knell for the notion of “junk DNA”. A moment’s thought counsels a critical response — for one thing, if the assertion is true, how can you account for the wellknown “C-value paradox”, whereby genome sizes can vary enormously among even closely related species [3]. The assertion in the media was admittedly an exaggeration of what the paper actually said, but given the considerable importance of the point and potential for misunderstanding, I think one could reasonably have expected that, following careful scrutiny by referees and editors, the authors would have been pressed to consider more carefully the biological meaning of the term “functional”, and of all the evidence supporting the contrary view (that eukaryotic genomes do generally have a large component of DNA that is “non-functional”, in the sense that it is not directly contributing to an organism’s evolutionary fitness). This should, one hopes, at least have reduced the chances of the kind of misrepresentation in the media that ensued. And here I think lies the rub: Nature clearly wanted to publish these papers and would I suspect justify allowing the clearly confusing claims by the value of the underlying substance — the ENCODE data. It might be better in such cases for the basic data to be published in some archival journal of the field, followed, if the researchers feel that those data can be used to support biologically thrilling claims, by a carefully argued attempt to do that in a devoted publication that is subject to an appropriate level of scrutiny prior to acceptance by any “high profile” journal. The second cause for worry is at first glance a small matter — a “Core Concept” article [4] written by a science writer for that august journal Proceedings of the National Academy of Sciences of the United States of America. The problems illustrated here are both specific and general. The general one is that the piece entirely lacks any appreciation of the historical context of the field — ostensibly “epigenetics”, but in the appropriately broad sense, gene regulation. The more specific one is the article’s complete failure to present a cogent account of the “concept” it is supposed to be explaining. I do not wish to be too hard on the author — the problems with the article also exist more broadly in the epigenetics literature — but surely one would expect better of PNAS. The article starts about by presenting the “core concept” of epigenetics as though it refers ineluctably to covalent modifications of DNA or histones. This view makes no sense (though it is not difficult to see how one might reach such a view from reading some of the “epigenetics literature”). I think everyone agrees that “epigenetics” concerns the way that a DNA genome can be expressed in different ways that are self-perpetuating, without changing the DNA sequence itself — this is what distinguishes it from “genetics”. This is a functional definition, and you cannot then reduce the term to a purely mechanistic one (at least, not unless there is definitive evidence to support such a one:one relationship). The question for the field must be: what mechanisms can result in distinct states of genome expression that are inherited, across either mitotic divisions (as during development), or meiotic divisions (as in trans-generational epigenetic effects)? We actually have one well-worked out “mechanism” for a clear epigenetic process: this is the lambda “epigenetic switch”, worked out over many years by Mark Ptashne [5] and colleagues in detailed studies which — building on the pioneering work of Jacob and Monod [6] more than 50 years ago — led to a beautiful picture of how transcriptional factors, working via feedbacks in regulatory “loops”, can stably maintain (epigenetic) states of gene expression across cell divisions. It is a constant cause of surprise to me that this paradigmatic work is not generally considered in papers that claim to address issues of epigenetics, particularly as the same underlying mechanisms have long been recognized by those studying development as lying at the heart of differentiation and pattern formation in higher eukaryotes. Both of these tales tell of “high profile” journals not clarifying biological progress for the broader scientific community and the outside world, but adding to confusion. There have in recent times been concerns expressed about problems with journals being run by “professional editors”. My contention is that these are wide of the mark — the first problem discussed above does admittedly concern a paper in a journal run by professional editors, but the second shows clearly that having a journal run by scientists does not, by any means, preclude poor judgement. But then we knew that, didn’t we.