Skip to Content

And Now, the Retractome

Back in January, I wrote about the controversial “Reactome” paper that had appeared in Science. This is the one that claimed to have immobilized over 1600 different kinds of biomolecules onto nanoparticles, and then used chemical means to set off a fluorescence assay when any protein recognized them. When actual organic chemists got a look at their scheme – something that apparently never happened during the review process – flags went up. As shown in that January post (and all over the chemical blogging world), the actual reactions looked, well, otherwordly.
Science was already backtracking within the first couple of months, and back in the summer, an institutional committee recommended that it be withdrawn. Since then, people have been waiting for the thunk of another shoe dropping, and now it’s landed: the entire paper has been retracted. (More at C&E News). The lead author, though, tells Nature that other people have been using his methods, as described, and that he’s still going to clear everything up.
I’m not sure how that’s going to happen, but I’ll be interested to see the attempt being made. The organic chemistry in the original paper was truly weird (and truly unworkable), and the whole concept of being able to whip up some complicated reactions schemes in the presence of a huge number of varied (and unprotected) molecules didn’t make sense. The whole thing sounded like a particularly arrogant molecular biologist’s idea of how synthetic chemistry should work: do it like a real biologist does! Sweeping boldly across the protein landscape, you just make them all work at the same time – haven’t you chemists every heard of microarrays? Of proteomics? Why won’t you people get with the times?
And the sorts of things that do work in modern biology would almost make you believe in that approach, until you look closely. Modern biology depends, though, on a wonderful legacy, a set of incredible tools bequeathed to us by billions of years of the most brutal product-development cycles imaginable (work or quite literally die). Organic chemistry, though, had no Aladdin’s cave of enzymes and exquisitely adapted chemistries to stumble into. We’ve had to work everything out ourselves. And although we’ve gotten pretty good at it, the actions of something like RNA polymerase still look like the works of angels in comparison.

14 comments on “And Now, the Retractome”

  1. Virgil says:

    What? A high profile paper in a high profile journal being retracted? Never! Well, at least they did it quickly (1 year).

  2. anchor says:

    Derek : back then when you reported, I read through this paper with due diligence and came away with a conclusion that retraction was the only course. The result presented were too good to be true and we all know what that meant!

  3. Frogs says:

    Next on the Science retraction agenda:
    Gaunt’s erroneous paper on “copper catalysed” Meta-Arylation.
    Someone teach this guy what a friedel-crafts reaction is!

  4. noname says:

    From the C&EN write-up: “Biochemist and Nobel Laureate Richard J. Roberts, chief scientific officer of New England Biolabs, thinks the decision to retract is premature.” Either he’s getting old and senile or he stumbled into his Nobel.
    Nothing about this paper is feasible, people. The errors in the figure were the absolute least of the problems. Go back and look at the discussion from Derek’s original post. The ignorance in the supp info was breathtaking. Retraction was the only option.

  5. G2 says:

    Nice that you have to pay to read the retraction – Science wants to make still more profit

  6. Carmen says:

    Thanks for the link, Derek. @noname The Scientist published an in-depth interview with Roberts back in August in case you’re interested in reading more on that take. Google “why the reactome is real” for the article.

  7. Spiny Norman says:

    “Modern biology depends, though, on a wonderful legacy, a set of incredible tools bequeathed to us by billions of years of the most brutal product-development cycles imaginable (work or quite literally die).”
    It’s an obvious thing to say, of course, but that’s just beautiful prose, Derek.

  8. I have to agree with Roberts. Retraction should not lead to detraction. The potential value of such methods seems significant, so other researcher should now take up where these guys left off instead of just ignoring the whole thing.

  9. ronathan richardson says:

    Roberts was supposed to do some double-blind experiments with samples from NEB on the reactome arrays, so maybe he saw something good on those? But this is of course the wrong way to test if the arrays are valid–they will undoubtedly give a mess of fluorescence changes when different samples are added, but so would a slurry of reactive dyes mixed with proteins on an array–it’s just not quantifiable science. In fact the definitive comment on this matter was given by “trisynthon” in the january thread:
    “I don’t doubt they actually did what they say. They probably did expose 1600 compounds to this iodinase enzyme and a dye and histidine. They probably did put that witches brew onto slides and the slides probably did look cool after cell lysates were added. It’s fun and cool, but it’s not science.”

  10. Zack says:

    It is incredible that the authors continue to claim that this works and only admit there were chemistry errors. The reactome array could not work, even if the chemistry was right. The idea is impossible.
    The problem is that the reactome array is based on the synthesis of a unique substrate linked to two dyes for every enzyme in the proteome. The enzyme cleaves the substrate, separating the dyes, producing a fluorescent signal.
    With this strategy, it is impossible to detect, even in theory:
    – Any reaction where one-half of the cleaved bond is a proton or hydride — no place to attach the dye. For the same reason you can’t detect hydration or dehydration, decarboxylation, deamination, dephosphorylation, etc…
    – Any isomerization (no cleaved sigma bond)
    – Any oxidation or reduction
    – Any reaction that involves a substrate that is metabolized by more than one enzyme in the cell (because then it is impossible to tell which enzyme did the reaction). For this reason, you cannot detect any reaction involving any cofactor such as: NADH, NADPH, ATP, GTP, SAM, TPP, acetyl-CoA, etc, etc…
    How serious are these limitations? For the following metabolic pathways, the reactome array could detect, in theory:
    Glycolysis: 1/10 reactions
    Krebs cycle: 0/10 reactions
    Pentose phosphate pathway: maybe 1/~10 reactions
    Fatty acid biosynthesis: 0 reactions
    Pick any metabolic pathway –almost every reaction is impossible to detect, in principle, with the reactome array strategy. Even the small number of reactions that are “possible” are totally implausible if you just look at the substrate and ask “Where should I attach the dye?”
    Obviously there other problems with the paper, such as: the chemistry is insane; the authors claim they can capture any enzyme with a cobalt bead, which is ludicrous; not a single control experiment was performed, etc… But even without these problems that everyone has noted, the approach is not logical, it cannot work.

  11. george p says:

    What should also be retracted is the job that this charlatan – Manuel Ferrer – continues to hold in an academic institution. The reactome paper was pure fabrication and nothing else – anyone with a basic knowledge of chemistry realizes that. Alas, one can be an editor of Science and be completely ignorant of elementary chemistry.

  12. noname says:

    thanks for the tip. I found this quote from Roberts to be particularly interesting:
    “The paper wasn’t about the chemistry; the paper was about what great biology you could do if you have this tool available to you.”
    By that standard, I should start preparing my manuscript “Tapping the Time-ome: A Practical Approach to Time Travel.” I haven’t worked out all the physics yet, but think what you could do if you had this tool available to you.

  13. trisynthon says:

    @ronathan: Always an honor to be quoted. May all your yields be 99%! (I’d believe you.)
    @Zack: It’s not inconceivable that one could design a substrate for a redox/dehydrating/etc. enzyme for which turnover would, rather than directly cleave the bond to the dye, at least render it labile. I remember seeing cascade fragmentation of dendrimers triggered by a single azide to amine reduction (anyone know this work?). The point is, each instance of design and confirmation of an enzyme/substrate pair would be a paper unto itself. Maybe not Science, but certainly Nat Chem Biol. But rather than publish 1600 papers in medium-impact journals, the authors elected to write one high-impact paper. Probably not the best decision in retrospect.

Comments are closed.