An ever-expanding methodological toolbox is one of the privileges of 21st century science. For most of us, the pace of Moore’s law-driven technological advancement is now so familiar as to be almost taken for granted. However, the mere availability of technologically-advanced new methods is not a guaranteed solution to any scientific problem. Thinking about the application of new approaches, ensuring that they genuinely build upon previous work (not to mention justify often hefty outlays from funders) is still an important challenge for scientists to take on – at least until the robots get really advanced.
Genomic science is, of course, no toddling newcomer any more – although methodological advances within the field continue to occur with head-spinning rapidity. But one aspect of the genomic ‘revolution’ that, for clear practical reasons, is only just beginning to emerge, is the availability of multi-generational genomic data.
For developmental scientists, this inevitable offshoot of, typically, the collision of increasingly affordable genotyping and existing multi-generational study samples, represents a potential goldmine of information and increased methodological rigour. Unravelling genetically-confounded associations between parent and child characteristics, or between child outcomes and their parent-provided/sibling-inhabited environments is a constant challenge in developmental research – and failure to do so (or, in some dispiriting cases, even attempt to do so) seriously undermines the interpretability of findings.
Widespread availability of genotype data for multiple members of the same family could change all this. But, as ever, the key part of the phrase ‘potential goldmine’ is the word potential. The mere availability of such tools is no guarantee that we will use them efficiently, informatively, or even for the right reasons. Novelty for novelty’s sake is a not a good use of research funds that will likely, for many of us at least, be increasingly hard to attract in the coming years (Trump, Brexit – I’m looking at you…).
Investigating intergenerational transmission: the current picture
Previous work aimed at investigating the mechanisms of phenotype transmission within families has made extensive use of biometric modelling techniques. Specifically, this involves the decomposition of cross-generational phenotypic associations in genetically-informed datasets using either adoption samples, classical child-as-twin samples, or Children-of-Twins (CoT) samples. By using estimates of genetic similarity derived from known relationships between individuals within families in these samples (e.g., 100% genetic similarity between identical twins, 50% between non-identical twins, siblings, and biological parent-child dyads, ≈0% similarity between adoptive children and parents), the relative extent to which phenotypic similarities are due to genes and environments can be inferred. The CoT design in particular, allows for not only the variance in a given phenotype to be decomposed, but also for intergenerational transmission to be partitioned into direct genetic and social transmission pathways. For more on CoT designs, and what they tend to find, see Tom McAdams’s recent explainer on just this topic.
“Irrespective of the strengths and weaknesses of current models, validating results obtained using them by applying different methodological approaches to the same questions is a core scientific principle”
Biometric designs have many advantages, but current samples may be under-powered to detect genetic transmission effects. Moreover, establishing the direction of effects between child and parent is not straightforward in such designs, with parent-driven social effects, child-driven social effects, and child-driven genetic effects largely remaining confounded. Irrespective of the strengths and weaknesses of current models, validating results obtained using them by applying different methodological approaches to the same questions is a core scientific principle. In behaviour genetics, one of the great successes of recent years has been the incorporation of genomic data, and accompanying development of methodological techniques, to seek validation of results from twin and family studies – resulting in a high degree of replicability for key findings 1.
Uses of multi-generational genomic data so far
As yet, very few published studies have incorporated genomic data into models of familial intergenerational transmission. In part, this is due the fact that multi-generational genomic data are currently available in very few samples. However, several genomic models and tools exist that are appropriate for adaption to developmental questions. One is Genome-wide Complex Trait Analysis (GCTA) 2 , a method that aggregates genetic similarity (in terms of concordance in measured single-nucleotide polymorphisms [SNPs] across the genome) in ostensibly unrelated individuals, and leverages it against phenotypic similarity in the manner of biometric models described above. One study has extended this method to examine maternal genetic effects on offspring traits 3.
A second genomic method that has begun to see use in a developmental context, is that of genome-wide polygenic scoring (GPS). Polygenic scores are individual-specific values based upon the number of “risk” alleles carried by an individual, as derived from summary statistics of large-scale genome-wide association studies (GWAS). The GPS approach aggregates the effects of individual SNPs from across the genome on complex traits such as depression, thus providing a flexible tool for phenotypic prediction from genotypic data. Polygenic scores have begun to be incorporated into investigations of intergenerational transmission only very recently, with just three published studies using them to try and distinguish genetic and environmental mechanisms 4–6.
Ultimately, the time that it will take for multi-generational genomic data to become widely available may offer us a rare chance to reflect on how to make the best use of them. Recent findings demonstrating limitations of polygenic approaches should be borne in mind – but so too should the potential of this emerging resource to invigorate the field of developmental science.
- Plomin, R., DeFries, J. C., Knopik, V. S. & Neiderhiser, J. M. Top 10 Replicated Findings From Behavioral Genetics. Perspect. Psychol. Sci. 11, 3–23 (2016).
- Yang, J., Lee, S. H., Goddard, M. E. & Visscher, P. M. GCTA: A tool for genome-wide complex trait analysis. Am. J. Hum. Genet. 88, 76–82 (2011).
- Eaves, L. J., Pourcain, B. S., Smith, G. D., York, T. P. & Evans, D. M. Resolving the Effects of Maternal and Offspring Genotype on Dyadic Outcomes in Genome Wide Complex Trait Analysis (‘M-GCTA’). Behav. Genet. 44, 445–455 (2014).
- Zhang, G. et al. Assessing the Causal Relationship of Maternal Height on Birth Size and Gestational Age at Birth: A Mendelian Randomization Analysis. PLoS Med. 12, 1–23 (2015).
- Richmond, R. C. et al. Using Genetic Variation to Explore the Causal Effect of Maternal Pregnancy Adiposity on Future Offspring Adiposity: A Mendelian Randomisation Study. PLOS Med. 14, e1002221 (2017).
- Marceau, K. et al. Passive rGE or Developmental Gene-Environment Cascade? An Investigation of the Role of Xenobiotic Metabolism Genes in the Association Between Smoke Exposure During Pregnancy and Child Birth Weight. Behav. Genet. 46, 365–377 (2016).