'The Most Horrific Time of My Career': When Years of Published Work Are Built on an Error

Leto Sapunar

January 14, 2021

In September 2019 Nicola Smith, a molecular pharmacologist in Australia, faced a brutal decision. She'd realized that she'd made a mistake — or rather, failed to catch a mistake in her group's research before the crippling error was published — in two academic articles which were the culmination of years of work. And she could either tell the world, or pretend it never happened.

Her students had been having trouble reproducing lab data. Once she looked into it and she figured out why, she told them, "Guys, you're not going to believe this." A cloning error had ensured the experiments were doomed to fail from the start.

If she came clean, she knew that at least one of the articles would most likely be retracted and she'd have to live with a lasting mark on her and her team's record. "What can I do to minimize the impact" on her two students? Smith thought at the time.

In particular, Tony Ngo,who was first author on both papers and had recently finished a PhD in her lab, was looking forward to a future in academia. Smith was terrified of tarnishing his prospects.

What was to stop her from just keeping quiet about it?

Smith, then at the Victor Chang Cardiac Research Institute in Sydney, studied G protein-coupled receptors – which are the largest class of receptors, are often targeted by drugs, and allow many hormones to interact with cells in the body.

She specialized in "orphan" G protein-coupled receptors, those activated by unknown molecular partners. At that time, she was looking into one labeled GPR37L1, which exists in humans and affects the cerebral development of mice. The research community studying the receptor is small and scientists are still trying to decipher its activity. "There is a lot of mystery around this receptor," says Irina Kufareva of the University of California, San Diego, who collaborated with Smith and was co-senior author on one of the studies.

In Smith's area of study, researchers can't observe a receptor directly. Instead, they use a genetic vector to carry it inside cells, then pay close attention to the resulting activity – looking for telltale chemical indicators of what the receptor might be doing.

In 2016, her group published a paper in Science Signaling in which they reported what happened when they cut off the head of the protein in question: the receptor's activity dropped – effectively switching activity on and off. They were able to show that the G protein-coupled receptor could be active without being triggered and could be controlled by severing that first chunk.

An indicator which helped them monitor this activity had also been present in their previous tests using yeast cells. Which was encouraging — now with human cells they were seeing similar results.

Then in 2019 an honors student (a post-undergraduate researcher) in Smith's lab was trying to make genetic constructs with mutated receptors to compare how they behave in cells with the natural ones. The constructs required for the experiments were comprised of a vector embedded with the genes of interest.

When he couldn't, they figured the failure was simply because the process was "fiddly," as James Coleman, then a post-doc in Smith's lab and co-first author on the paper with Ngo, puts it. The trouble persisted until August 2019, when they gave up and decided to order ready-made versions of the constructs instead.

But when the constructs arrived, and the lab tried to recreate the past experiments, the new construct "was behaving completely unlike the one that we had been working with previously – the one that we had published with," Coleman says.

In September they sequenced the troublesome genes and found they had been using the wrong vector to carry them.

That wasn't all. Not only was the gene put in the wrong vector, the receptor gene itself — a string of genetic information — was inserted into the vector backwards so it couldn't be read correctly. Even one of those catastrophic errors would have almost certainly invalidated her findings, but both together killed all chances of a meaningful result.

The reason it was so hard to catch the snafu sooner was that normally when one makes a mistake like this, Smith says, the experiment simply wouldn't work. Remarkably, not only had the experiment worked, it had produced results that were in line with what the researchers had seen in related experiments. Now that the team had arrived at the root of the problem, their hearts sank as it became clear that their once-promising results were a fluke.

After realizing her group's mistake, Smith turned to a trusted colleague for advice. She says the colleague told her that "despite the fact that you really care about this receptor, no one else in the world really gives a toss about it."

In essence, the sentiment was: Weigh the damage done by letting an error affect a few niche publications against the potential fallout and long-term career damage it would do to her and her team. And for that day Smith was almost convinced that keeping quiet to protect her team was the best option.

After hours of turmoil, Smith realized that her colleague's line of reasoning – that she could ignore her error and just move on – was "utter bullshit," she says. It didn't matter whether or not anyone else cared about the receptor; feigning ignorance was wrong.

You just have to do the right thing, Smith says, "even though it's the most painful thing you'll ever do." She couldn't spend the rest of her career wondering how much damage her error would cause. She alerted her institutes about the error (both Victor Chang and the University of New South Wales, with which it's affiliated) and, shortly after, the journals involved. She decided that total transparency was the best path.

After making that choice, "The weight just lifted off my shoulders; it was the right decision," Smith says. She was committed to doing the right thing, but knew she'd have to face consequences.

Months after Smith's realization of the error, she found herself in the heart of a storm. She was immensely stressed, getting regular phone calls from the investigators at her institute, whose funding body required them to launch a preliminary investigation to see if there were signs of misconduct, Smith says.

She was organizing experiments for her students to contextualize the error. One of those students was Brendan Wilkins, a research assistant who often worked until midnight.

Smith did all of this while taking care of her two boys, one age three, the other not yet one. The ordeal had pulled her away from her family and began not long after she returned from maternity leave. "[M]y three-year-old started mimicking me at the computer, saying: 'Sorry, Mummy, I'm doing work,'" because it was so all consuming, Smith says.

But at the worst of it, she got an email from another researcher which strengthened her resolve. The researcher had a question about the Science Signaling paper, which she and her co-authors had asked the journal to retract. It would be, but at the time the forward facing text was unchanged. That moment drove home to her the importance of correcting the record. Scientists spend so much time and energy struggling to reproduce genuine data, Smith says, there isn't room in the literature for known errors.

Smith explained what had happened and asked the researcher to be discreet while the retraction was being worked out.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.