By Jey McCreight, Ph.D., Research Communications Scientist
It takes a village to do research. 23andMe’s scientific discoveries are made possible both by our more than three million customers who’ve consented to participate in research, and by our collaborators. Those collaborators include people like Varun Warrier, MPhil, a graduate student in the laboratory of Simon Baron-Cohen at the University of Cambridge. Warrier has lead a number of collaborative studies with 23andMe, including most recently two studies on empathy published in Translational Psychiatry and Molecular Psychiatry, and another on “theory of mind” published in Scientific Reports.
JM: What motivated you to first start studying autism?
VW: I don’t think there was anything specifically that motivated me to study autism. I was interested in genetics and neuroscience, and during my undergraduate and my master’s, I tried to choose between the two of them. Then, at some point I realized that I didn’t have to choose — I could use methods in genetics to investigate the biology of mental health conditions. I wrote to Simon and asked if I could join his lab to do this research. What really attracted me to the Autism Research Centre was the fact that they were looking at the dimensional aspect of autism, which, to me, was a lot more exciting than thinking of mental health conditions as existing in silos.
JM: How has the rapidly increasing availability of large datasets changed the field of psychiatric genomics?
VW: The progress has been immense. The earliest genome-wide association study (GWAS) of a psychiatric condition was about a decade ago. Until about 2013 or so, we had a handful of genetic variants associated with any psychiatric condition or related traits. In the last five years, thanks to large datasets like 23andMe and the UK Biobank, the number of variants discovered has seen a massive jump. For instance, just about a month ago, a preprint was posted on BioRxiv that identified more than 600 genetic regions associated with risky behavior. We’ve crossed a milestone in psychiatric genetics. The next challenge is to link this to biology and use these polygenic scores to inform diagnosis and interventions.
JM: What are the benefits of using GWAS to investigate behavioral traits like autism? Are there any limitations?
VW: First, the limitations. The underlying risk for developing autism is only partly genetic (60 to 80 percent). With traits like empathy or theory of mind, genetics account for an even smaller proportion of the variance. So we are very much working within that fraction of variance when we look at genetics. The other thing to bear in mind is that the environment plays a vital role in whether someone chooses to receive an diagnosis or not. For instance, if you’re in a supportive environment that allows you to flourish and have high autistic traits, then you may choose not to receive a diagnosis. There are quite a few other factors besides genetics that determine the relative wellbeing of an individual.
That said, the benefits far outweigh the limitations. Personally, I think trying to delineate the genetic mechanisms underlying autism is vital for us to understand anything else about the condition. The interesting thing about genetics is that, by and large, we can assume that reverse causality is absent. With almost anything else, you aren’t entirely sure if what you’re observing is a cause or a consequence of the condition. Once you better understand the genetic mechanisms of a condition like autism, you can use that to better understand things like neural circuits or methylation signatures in autism and understand causality.
JM: Your recent papers investigate different processes that contribute to autism, such as
theory of mind, empathy, systemizing, and relationship satisfaction. Why is it important ? to look at these individual processes, rather than autism as a whole?
VW: Autism is really an umbrella term to refer to a heterogeneous group of conditions. We can think of it in two ways. In the first scenario, we can go by the diagnostic label of autism for research. But we know that there is immense heterogeneity within the condition, which we may not really be able to dissect genetically unless we look at it from other perspectives. An alternative is to investigate traits that are normally distributed in the general population but in which autistic individuals, on average, have strengths or difficulties in — or in other words, thinking of autism as being along multiple different dimensions. Our lab has a longstanding interest in investigating the dimensional nature of autism, and thanks to 23andMe and other genetic biobanks like the UK Biobank and ALSPAC, we were able to do just that using genetics.
JM: You recently identified three loci significantly associated with systemizing, a non-social autistic trait. What is systemizing, and why is it important to consider both social and non-social traits when investigating the genetic architecture of autism?
VW: Systemizing is an interest in or a talent for analyzing or building systems. If we think of all the information around us, we can try and classify them or order or group them based on some intrinsic properties in the natural world. For example, think of colors. When we think of colors we don’t necessarily go and classify them. But if we wanted to, we can order them based on the natural visible spectrum. We could group them into warm or cold colors. We could order them based on brightness. As scientists, we like to find order and pattern in things to better understand them. The same seems to be true, on average, with autistic individuals. Research from our lab and others have demonstrated that, on average, autistic individuals have superior interests in systems. While most people think of autism as a social condition, there is also a non-social aspect to it that’s necessary for a diagnosis — unusually restricted or repetitive patterns of interest and behavior. We think that it is aspects of these that are reflected in the superior systemizing skills that are frequently observed in autistic individuals.
JM: Two of your recent GWAS did not find any significant loci associated with “empathy” or “theory of mind” (the ability to attribute mental states to one self and others and to use such mental state attribution to make sense of behaviour and predict it). Why do you think this is so?
VW: This essentially boils down to sample size. We know that there is a genetic signal in these GWAS. Differences in the theory of mind and empathy are heritable. We know this from twin studies that show that approximately a third of the differences in how empathetic people are is due to genetics. Our studies have also identified a small but significant additive SNP heritability. But the variance explained by each individual SNP is small. To find these really small effects, we need large sample sizes.
JM: Your studies suggest shared genetic architecture between different measures of theory of mind and cognition, as well as between empathy and autism, schizophrenia, and anorexia nervosa. Why are psychological traits so interconnected? Does this present any challenges or opportunities for your research?
VW: We can think of any psychiatric condition or cognitive and behavioral trait as being emergent properties of neural networks. Genes contribute to these phenotypes primarily by altering these neural networks. They alter development, synaptic signaling, glial function and so on, all of which contribute to altered information processing in the brain. If a handful of these neural networks are altered, they will contribute to differences in multiple phenotypes, which probably explains the shared genetics that we see between all these different conditions and behaviors. It’s challenging in that we need to figure out why some people get diagnosed with autism and some others get diagnosed with ADHD when we know that there is considerable overlap in common and rare variants associated with the conditions. It’s a lot more tricky that anticipated, and we will have to investigate both interactions between genetic variants and interactions between genetic variants and other factors to better predict the differences in these conditions. On the other hand, we can leverage the shared genetics between multiple conditions to increase the statistical power in our studies and identify genetic variants.
JM: Across the number of autistic traits you measure, women consistently score as less autistic than men. Why do you think we see this sex bias in autism?
VW: We think there are multiple reasons for the sex bias. Some of it is biological. We know that women diagnosed with autism have, on average, a larger number of rare, usually de novo, protein truncating variants associated with autism compared to men diagnosed with autism. We have some evidence for sex-differences in the genetic variants that contribute to some of the traits associated with autism such as empathy. There could be other biological reasons — studies have demonstrated that elevated prenatal androgens are associated with increased risk for autism.
There are also other non-biological factors. Women are less likely to be diagnosed with autism as many clinicians may think that autism is primarily a condition seen in men. There is evidence that autism manifests differently in women. Women often need more autistic traits than men to receive a diagnosis. Finally, the diagnostic criteria for autism may not recognize how autism manifests in women. We need to explore all these factors, and it’s likely that both the biological and social factors contribute to the sex bias in autism.
JM: What made you first interested in collaborating with 23andMe?
VW: 23andMe had approached the Autism Research Centre a few years ago to include some of the tests developed by researchers there in the customer website. At some point in 2012, when the early GWAS studies on autism were published, we realized that we need to investigate the dimensional nature of autism. We reached out to 23andMe to see if they’d be happy to collaborate once again on these phenotypes and conduct a GWAS.
JM: What’s your opinion now after collaborating with 23andMe?
VW: We’ve had a series of fruitful collaborations with 23andMe. I remain highly positive in general of collaborations with a consumer genetics company.
JM: Your advisor Simon Baron-Cohen has developed autism support services such as
The Transporters, an animated DVD series that teaches emotion recognition to children with autism. Does your lab have any further plans to develop autism support services based on your research?
VW: Yes, our lab is conducting a series of randomized control trials on various interventions, including apps that can help minimally verbal autistic children.
JM: What other research projects do you have planned for the future?
VW: We want to continue our work on the genetics of dimensional measures associated with autism. In particular, I am interested in seeing if this can be used to stratify individuals within the autism spectrum.