More Recent Comments

Thursday, January 16, 2014

Press release hyperbole and the "duon" delusion

I recently described a really bad paper published in Science by Stergachis et al. (2013). The principle investigator is John Stamatoyannopoulos of ENCODE notoriety [see The "duon" delusion and why transcription factors MUST bind non-functionally to exon sequences].

The group mapped millions of transcription factor binding sites in the human genome and discovered that 1.8% of them were in exons (coding regions). They assumed that these were functional—they play a role in regulating gene expression. Thus the nucleotide binding sites are also codons meaning that the sequence specifies two different kinds of information. The workers named these sequences "duons."

I pointed out in my blog post that most of these transcription factor binding sites had to be spurious binding sites with no biological function. This is not a new idea. It was well-known even before the ENCODE publicity fiasco of last year but in the past 15 months there has been a lot written about the properties of DNA binding proteins. The ENCODE claim that all of them are biologically relevant has been challenged repeatedly.

It's okay if you want to believe that all of these bound transcription factors are doing something useful even if it's completely contrary to everything we know about the biochemistry of DNA binding proteins and the organization of genomes. It's not okay to completely ignore the possibility that you may be looking at something that's not biologically relevant. That's bad science.

Here's the press release from the University of Washington; Scientists discover double meaning in genetic code.
Scientists have discovered a second code hiding within DNA. This second code contains information that changes how scientists read the instructions contained in DNA and interpret mutations to make sense of health and disease.
Ugh! That's totally absurd for many reasons. First, the idea that the DNA sequence in exons may serve another function has been known for over thirty years (3′ splice sites, sites where RNA processing enzymes bind) (see figure below). Second, DNA binding sites are not a "second code." Third, the authors of the paper did not show that their binding sites had a function.
Since the genetic code was deciphered in the 1960s, scientists have assumed that it was used exclusively to write information about proteins. UW scientists were stunned to discover that genomes use the genetic code to write two separate languages. One describes how proteins are made, and the other instructs the cell on how genes are controlled. One language is written on top of the other, which is why the second language remained hidden for so long.
Does anyone honestly think that Stamatoyannopoulos and the people in his lab were "stunned"? It makes them sound awfully stupid.
"For over 40 years we have assumed that DNA changes affecting the genetic code solely impact how proteins are made," said Stamatoyannopoulos. "Now we know that this basic assumption about reading the human genome missed half of the picture. These new findings highlight that DNA is an incredibly powerful information storage device, which nature has fully exploited in unexpected ways."
Hmmm ... maybe they really are stupid after all.

Emily Willingham on the Forbes website makes a good point about the dangers of such hyperbole in university press releases [Don't Be Duped By 'Duon' DNA Hype].
The hype began with the way hype often begins: an institutional news release offering us the holy grail/huge breakthrough/game-changing finding of the day. This kind of exaggeration is the big reason any science consumer should look well beyond the news release in considering new findings. A news release is a marketing tool. You’re reading an advertisement when you read a news release. In this case, the advertisement/news release not only goes off the rails with the hype, it’s also scientifically garbled and open to all kinds of misinterpretation, as the comments at the link to the release make clear.
I agree that this is a serious problem. One might naively think that a university press release would be scientifically accurate and would faithfully describe what's in the published paper. Unfortunately, that's not true anymore (if it ever was true). The people who write press releases don't read the paper. They rely entirely on what they are told by the authors of the work. And those authors are inclined to add "embellishment" to make their paper sound a lot more important that it really is.

I don't know how to fix this problem but I do have one suggestion. Scientists often excuse what's in the press release on the grounds that they didn't write it and their views aren't being properly represented. I think that every press release should be authorized by the university researchers in the same way that political messages have to be authorized by the candidate. The bottom of each university press release should contain a statement like: "The following authors ..... have read and approved this press release. They affirm that the information presented here is a fair and accurate representation of the results and conclusions published in the peer-reviewed paper."

So much for university press releases. Now let's look at another kind of press release. The major journals often advertise important papers by writing up a short summary in the front of the journal. In the case of Science this is the section called "Perspectives." Sometimes these summaries are written by journal staff writers and sometimes they are written by scientists who are solicited by the journal.

In this case, the summary was written by Robert Weatheritt and Madan Babu at the MRC Laboratory of Molecular Biology in Cambridge, UK. The title is: The Hidden Codes That Shape Protein Evolution.

Now, you might think that the review in Science would be better than a university press release. If so, you would be right. But there's good news and bad news. Here's what Weatheritt and Babu say,
Despite redundancy in the genetic code, the choice of codons used is highly biased in some proteins, suggesting that additional constraints operate in certain protein-coding regions of the genome. This suggests that the preference for particular codons, and therefore amino acids in specific regions of the protein, is often determined by factors unrelated to protein structure or function. On page 1367 in this issue, Stergachis et al. reveal that transcription factors bind within protein-coding regions (in addition to nearby noncoding regions) in a large number of human genes. Thus, a transcription factor "binding code" may influence codon choice and, consequently, protein evolution. This "binding" code joins other "regulatory" codes that govern chromatin organization, enhancers, mRNA structure, mRNA splicing, microRNA target sites, translational efficiency, and cotranslational folding, all of which have been proposed to constrain codon choice, and thus protein evolution (see the figure).
The good news is that Weatheritt and Babu are smart enough to realize that codon constraints is old news. There are already lots of examples of codons serving a dual function like binding sites for factors that bind to RNA. This wouldn't be very remarkable except that Stamatoyannopoulos was quoted in the press release as saying, "For over 40 years we have assumed that DNA changes affecting the genetic code solely impact how proteins are made."

The bad news is that Weatheritt and Babu fall hook-line-and-sinker for the claim that transcription factor binding sites are functional rather than fortuitous accidents. They then go on to discuss the implications of evolutionary constraints on codon usage due to transcription factor binding. If the binding sites are non-functional then there aren't any evolutionary constraints and the entire point is moot. The reviewers don't seem to be the least bit skeptical about the claims of Stergachis et al.

That's a bit of a surprise but perhaps not as shocking as you might think. After all, none of the reviewers of the paper raised any questions so why would the scientists who write up the summary for the journal? (It's not unheard of for the people who reviewed the paper prior to publication to also write the commentary.)

All in all, I think this paper is a bit of a disaster. The university press release is ridiculous and embarrassing. The paper itself doesn't acknowledge the controversy over the functionality of DNA binding sites.1 And, the scientists who wrote up the advertisement for the journal were somewhat naive.

1. It's just not possible that Stegachis et al. were unaware of the controversy. When Stamatoyannopoulos was here last month he certainly knew that there were scientists who challenge the ENCODE conclusions. This means that the authors of the paper were deliberately suppressing information that weakens their case. That's not how we're supposed to do science.

Stergachis, A.B., Haugen, E., Shafer, A., Fu, W., Vernot, B., Reynolds, A., Raubitschek, A., Ziegler, S., LeProust, E.M., Akey, J.M. and Stamatoyannopoulos, J.A. (2013) Exonic Transcription Factor Binding Directs Codon Choice and Affects Protein Evolution. Science 342:1367-1372. [doi: 10.1126/science.1243490] [Abstract] [PDF]


Diogenes said...

I will repost my glossary used to describe such disasters.

Paradigm Shaft: lying about the content of a theory, and replacing it with a different, dumber theory, which one then claims to have disproven, inducing a revolutionary Kuhnian "Paradigm Shift."

Pressiopath: aka Press Release Sociopath; one who says a quotidian thing in his published paper, but claims to have made a revolutionary breakthrough in his Press Release, and has an untroubled conscience about lying to the non-scientific public.

Pubwhore: one who will do anything with his mouth in order get attention from the media.

Passive Tense Pussy: one who uses the passive tense for purposes of evasion, to avoid naming unknown scientists who are accused of having expressed some dumb in an unspecified text at an unspecified time; e.g. "long dismissed as Junk DNA...".

cf. Axel Visel: "Enhancers are part of the 98 per cent of the human genome that is non-coding DNA – long thought of as 'junk DNA'," says Visel. "It's increasingly clear that important functions are embedded in this 'junk'."

Marcoli said...

On the subject of codon bias, is it possible that bias can also be to improve fidelity of tRNA binding? I know there is 'wobble' in anticodon-codon binding, but might there be codons that are preferred b/c they are less wobbly?

SRM said...

Meanwhile, for those who prefer to work on specific questions of structure and function rather than high-throughput global analyses...if we were to make claims about the importance of a binding site without testing the consequence of binding experimentally, we would surely encounter a great deal of deserved flak during the review process.

Anonymous said...

Yup. There's been lots of articles about codon bias tending to correlate with tRNA versions, copy numbers, and such.

Marcoli said...

On a possibly bright note, a Google search for duon lists pictures of the mechanical thingy first, followed by several links that crap all over the idea of duons in the genome. I choose to interpret this as a good trend in how this is going.

caynazzo said...

"The people who write press releases don't read the paper. They rely entirely on what they are told by the authors of the work. And those authors are inclined to add "embellishment" to make their paper sound a lot more important that it really is."

Speaking from experience with University of Washington press release writers this, if anything, is an under exaggeration of the uninformed hack work they routinely produce. These press releases go out without the approval and oftentimes knowledge of the scientists whose work they're writing about.

caynazzo said...

Are you saying for the under appreciated yet superior few of us who only do functional work that if we didn't demonstrate that very thing which we claim to be all about we'd get ridiculed?