DNA, colour perception, and ‘The Dress’

Did you see ‘blue and black’ or ‘white and gold’? Or did you miss the ‘dress-capade’ that exploded the Internet last month? It was started by this post on Tumblr that went viral.

Many people warned their heads risked exploding in disbelief. How could people see the same dress in different colours?

It appears the variation lies in the way we judge how light reflects off objects of different colours, as Harvard psychologist Steven Pinker explained in Forbes.

A follow-on, calmer discussion started about whether this trait could be in our DNA.

Michael Whitehead made an open call for 23andMe, the consumer genetics company, to test whether DNA might be responsible.

Giovanni Marco Dall’Olio opened a (private) thread inside the 23andMe discussion forum directly asking for users to record what they saw. At last check, more than 500 people had posted responses.

Indeed, 23andMe did launch an online survey.

This prompted discussions started by Carl Zimmer on Twitter. One scientist reported she saw ‘blue and black’ while her son saw ‘white and gold’. Another saw ‘blue and black’ while his identical twin saw ‘white and gold’.

Pinker said his intuition was that it was not genetic, but he’d be keen to see the data.

The first question to answer is this debate is whether the trait is heritable.

Darwin knew about inheritance, but not its mechanism. It wouldn’t be until the turn of the 20^th century that Mendel’s experiments with sweet peas led to an understanding of how inheritance works and until the Avery-MacLeod-McCarty (1944) before it was known that genes resided in our DNA, the structure of which was cracked by Watson and Crick (1953) opening up the era of molecular biology.

Genomes aren’t needed to ask if a trait is heritable – just information from pedigrees. With his advanced undergraduate class in genetics, Steve Mount analyzed data crowdsourced from 28 families which showed no trend for the perceptions of parents to predict those of their children. The data did show some tendency for daughters to see the dress as fathers. Mount’s conclusion was that the trait can’t be strictly genetic, but could have an X-linked genetic component.

23andMe then released the results of their far larger study as a blog article and a more detailed whitepaper. Amazingly, in a few short days, 23andMe had convinced more than 25,000 of their customers to complete #TheDress survey. Their data confirmed the lack of any clear genetic link with colour perceptions of the dress. Yet, it did reveal a potential association with a variant of the gene ANO6, a gene in the same family as ANO2, which is involved in light perception that might be significant in a larger sample size.

Rather than being down to DNA, the strongest determinant was age followed by where you grew up. You are more likely to see ‘white and gold’ as you approach the age of 60. Those who grew up in more urban areas also saw ‘white and gold’ in higher proportions.

23andMe suggest this work warrants further study, especially as it underlines the complex interactions of genetics and environment (the Nature versus Nuture debate).

The lasting significance of this wonderfully geeky chapter of ‘the dress’ debate is how quickly data was leveraged to sate scientific curiosity. A question was posed and answered in short order. It marks a turning point in our thinking about how society can mobilize genomic data. Cached data is just waiting to have questions asked of it; 23andMe have more than 900,000 customer DNA profiles to hand, 80 percent of whom consent to participate in research such as this survey.

Dall’Olio wrote in his 23andMe post that “if we are fast and we collect a good number of answers, this can be the fastest discovery of association between a trait and its genotype in the history of genetics”. His comment was prescient.

The catch is combining the DNA with information about individuals. Who knew to ask people before-hand when they signed up what colour that dress was? Absurd to think anyone ever would.

Another issue is that 23andMe examines a significant proportion of the variation in the human gene pool but not ‘the entire genome’. There could be genetic variation for particular traits that go undetected.

In an ideal scenario, the perfect genetic study of a heritable trait would involve a large cohort of individuals with Trait A and Trait B and the use whole genome sequencing. But no one would fund such a ‘Dress DNA’ study. Curing cancer is far higher up the priorities list. Thus, existing databases of millions of genomes open up new possibilities for discoveries that might otherwise never be made.

A current roadblock lying on the path towards this emerging vision of true ‘community genomics’ where the curious public can mine our collective DNA, is that only 23andMe have access to customer profiles. 23andMe contribute to peer-reviewed papers and have numerous academic collaborations, but beyond these agreed uses of their data, scientists and the general public alike can only play on the sidelines.

What if such data were opened up? Everyone, has access to data contributed to the openSNP genome sharing platform. The catch with openSNP is that the number of people willing to make their DNA public is still small, but growing.

Whitehead created a survey for the phenotype for perception of ‘The Dress’. The first user to record his phenotype was Bastian Greshake, the creator of the openSNP platform. Appropriately, the growing list of users is split on ‘blue and black’ and ‘white and gold’.

Feature image credit: Colours in paper, by Soorelis. Public domain via Pixabay.