by Matthew Cobb
You may recall that back in October we reported the amazing discovery that, as I put it in the headline, “Neanderthal genes are everywhere“. Up until then, it had been thought that only those human populations outside of Africa carried Neanderthal genes, as a consequence of our ancestors having mated with our Neanderthal cousins—mainly in Europe and the Middle East. People from sub-Saharan Africa, it was thought, did not carry those genes, because their ancestors did not leave Africa, and so didn’t meet the Neanderthals (whose ancestors had left Africa several hundred thousand years earlier).
What happened in October was that a group of researchers from around the world, led by Gallego Llorente of Cambridge University, studied the DNA of Mota, an Ethopian man who lived around 4,500 years ago. They found that he carried an unexpectedly high proportion of DNA from European populations, including DNA that had originally come from Neanderthals over 30,000 years earlier (Neanderthals went extinct around 35,000 years ago). The really exciting bit came next. As I wrote:
When they compared Mota’s DNA with those of modern African populations, they found that the European sequences he carried were also present deep in the continent, even amongst the Pygmies of the Congo. Even ‘reference’ African genomes, such as those from the Yoruba and Mtubi peoples, which were thought not to have been affected by interbreeding with Europeans, turned out to have around 6% of their DNA from European DNA, like Mota. (…)
The final novelty came when the researchers looked at Neanderthal DNA. Mota carried that DNA, just like me, because his ancestors had mated with Neanderthals tens of thousands of years earlier.
And it turned out that some of those Neanderthal sequences could also now be detected in African populations, too. They were very dilute – around 0.5% – but they were clearly there. They do not indicate that there were Neanderthals in Africa, but rather that when the offspring of Mota and others carrying migratory European sequences spread their DNA into Africa, they also spread small amounts of Neanderthal DNA, too.
This was amazing, and I excitedly changed the lecture I gave to my first-year students the next day. I’ve since explained in various public lectures that we all carry Neanderthal DNA, including those of Afro-Caribbean origin.
Now it turns out not to be true.
The authors have published an erratum notice to explain the error, along with updated versions of the figures and tables. If you want the technical explanation, here it is:
A script necessary to convert the input produced by samtools v0.1.19 to be compatible with PLINK was not run when merging the ancient genome, Mota, with the contemporary populations SNP panel, leading to homozygote positions to the human reference genome being dropped as ‘missing data’
What this means is that they forgot to run a particular computer programme (‘script’) that would harmonise the outputs of two different programmes used to do the analyses of Mota and of modern sub-Saharan populations. When they did so, much of their effect disappeared.
As Ewan Callaway explains, the error came to light when two researchers, Pontus Skoglund and David Reich, tried to replicate the finding, doing their own analysis of the Mota genome. They failed, and alerted the original authors, who soon worked out their mistake. In an exemplary act of clarification, they let everyone know their error and have corrected their data.
The conclusion is that although there was migration back into Africa (what is delightfully called ‘back-flow’), this was less extensive in geographical terms than the authors claimed. The European genes carried by Mota did not flood all the way back down to Africa, and Modern sub-Saharan populations do not have unexpectedly high levels of European, and therefore Neanderthal, DNA. We are not all Neanderthals.
These two figures show the difference in the analyses. The first is from the October 2015 article, the second, corrected, version is from the Erratum:
The lessons of this are multiple. Above all, hats off to Llorente et al, the original researchers, for making their data openly available, so that quizzical scientists like Skoglund and Reich could explore what they thought to be an unlikely result. And then even more kudos for publishing their correction so quickly. This is how science works – if something seems weird, and it is incorrect, it should get corrected by science’s inherent questioning, based on experimentation.
There is another lesson, too. A lot of modern research is based on complex analyses that can be difficult for those outside of the research group involved to understand. A lot is therefore taken on trust; had the result not been unusual, the mistake might never have been discovered. I have experienced a similar problem when some exciting results we had found eventually turned out to be an artifact, produced by an error in a computer script. Although the work has not been published, so we did not need to retract, we probably wasted about 18 months getting excited about something that turned out not to be there. . .
Finally – what about my students? I told them about the Mota paper, and how they all had Neanderthal genes, even those of Afro-Caribbean origin. I’m going to change the lecture for next year, obviously, but my students – many of whom will not take another course in human evolution – may never discover that what I told them was wrong. And what about next year’s students? Do I tell them the whole story as an example of how science proceeds? If so, I know from bitter experience that for a sizeable number of them, the thing that will stick in their memory will be the wrong result, not the scientific lesson. I think I’ll keep quiet in the lecture, but post links to this and the original post. I’ll also edit the original post, with a pointer to this correction.
