SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

This selective review considers findings in genetic research that have shed light on how genes operate across development. We will address the question of whether the child is ‘father of the Man’ from a genetic perspective. In other words, do the same genetic influences affect the same traits across development? Using a ‘taster menu’ approach and prioritizing newer findings on cognitive and behavioral traits, examples from the following genetic disciplines will be discussed: (a) developmental quantitative genetics (such as longitudinal twin studies), (b) neurodevelopmental genetic syndromes with known genetic causes (such as Williams syndrome), (c) developmental candidate gene studies (such as those that link infant and adult populations), (d) developmental genome-wide association studies (GWAS), and (e) DNA resequencing. Evidence presented here suggests that there is considerable genetic stability of cognitive and behavioral traits across development, but there is also evidence for genetic change. Quantitative genetic studies have a long history of assessing genetic continuity and change across development. It is now time for the newer, more technology-enabled fields such as GWAS and DNA resequencing also to take on board the dynamic nature of human behavior.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

The quotation referred to in the title of this paper, ‘The Child is father of the Man’, from a poem by William Wordsworth (1770–1850), is interpreted by some scholars to say that our early life shapes whom we become as adults. We now know that early life is influenced by both our genes and our environmental experiences. In this selective review, evidence for the stability of genetic influences on early and later development will be considered. To what extent is the child ‘father of the Man’ in terms of genetic influences? It is important to note that the review focuses on common, complex cognitive and behavioral traits, on which it is believed that genetic influences act probabilistically rather than deterministically. In other words, to what extent do the same (probabilistically acting) genes influence similar kinds of individual differences in infancy, childhood, adolescence and adulthood? Without denying the importance of environmental influences, they are outside the scope of this review, and the reader is directed elsewhere for informative reviews on the processes that could underlie gene–environment interaction, including epigenetics and gene expression (Gottlieb, 2007; Kan, Ploeger, Raijmakers, Dolan & van der Maas, 2010; Plomin & Schalkwyk, 2007; Rutter, 2007; Westermann, Mareschal, Johnson, Sirois, Spratling & Thomas, 2007).

Why is this question important? First, it has been of long-standing interest to understand what causes individual differences in human behavior. Development is central to human behavior and as such exploring the causes of age-to-age change as well as continuity is an intrinsic part of understanding individual differences. Second, prevention is now a key priority in medical research (Sahakian, Malloch & Kennard, 2010). It is critical to know not just which genes are associated with a complex disease or disorder, but which genes are associated with the pre-disease, at-risk, prodromal state. Pre-disease, at-risk states are better targets for prevention than the disease once it has started. In this vein, infants at risk of (nonsyndromic) autism spectrum conditions are now the focus of much research, for example (Elsabbagh & Johnson, 2010; Noland, Steven Reznick, Stone, Walden & Sheridan, 2010; Young, Merin, Rogers & Ozonoff, 2009). An understanding of how genetic influences operate across developmental time would constitute an important knowledge base from which to develop preventative strategies for heritable but not genetically predetermined cognitive or behaviourally defined disorders.

Using a ‘taster menu’ approach, that is, picking particular examples from the field and prioritizing newer findings, examples from the following genetic disciplines will be discussed: (a) developmental quantitative genetics, (b) neurodevelopmental genetic syndromes with a known genetic cause, (c) developmental candidate gene studies, (d) developmental genome-wide association studies, and (e) DNA resequencing. The review focuses on findings that relate to genetic influences on the development of cognitive and behavioral traits.

Developmental quantitative genetics

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

Quantitative genetics is the investigation of genetic and environmental influences on complex traits and disorders (Plomin, DeFries, McClearn & McGuffin, 2008). The classic twin design is one of the main methods employed in quantitative genetics; it involves the comparison of the within-pair similarity of identical (or monozygotic) and fraternal (or dizygotic) twins. The twin design is equally informative about environmental influences as about genetics, but here we will focus on the genetic findings from twin studies.

The twin design is a particularly useful tool for studying development. Longitudinal studies of identical and fraternal twins studied at multiple ages can be informative about genetic continuity– the degree to which genetic influences on a trait or disease are stable across ages – as well as about two types of genetic change, genetic innovation– the degree to which new genetic influences are present at later ages – and genetic attenuation– the degree to which genetic influences present at earlier ages decline in influence at later ages (Kendler, Gardner, Annas, Neale, Eaves & Lichtenstein, 2008).

The basis of a longitudinal twin analysis is to compare the degree to which one twin’s trait score at the earlier age correlates with their co-twin’s trait score at the later age (called a cross-age cross-twin correlation), separately for identical and fraternal twins. If identical twins show a greater cross-age cross-twin correlation than fraternal twins, this suggests that the same genetic influences are operating on the trait across development. The model can also estimate if there are different genetic effects on the trait at different ages. A cross-age genetic correlation (rg) is a direct estimate derived from twin models of the degree of shared genetic influences across ages. This genetic correlation can vary from 0, indicating that completely different genetic influences play a role at two ages, to 1, indicating complete overlap in genetic influences at two ages.

In terms of cognitive development, a recent example comes from a study of 8700 twin pairs who were assessed from age 2 to age 10 years on their cognitive abilities. Across this 8-year age range spanning early to middle childhood, the authors found considerable stability in genetic influences (cross-age rg = 0.57) but also significant change (Davis, Haworth & Plomin, 2009). A second recent twin study of cognitive development employed two complementary twin cohorts who were assessed at different ages on verbal abilities (Hoekstra, Bartels, van Leeuwen & Boomsma, 2009). They showed that verbal abilities have a similar genetic architecture in childhood (9-year-olds) and late adolescence (18-year-olds), that is, the degree of genetic overlap between different subdomains of verbal ability was similar across the two ages. This careful matching of independent twin samples, while not providing a direct estimate of the degree to which the same genetic influences are operating at the two ages (because cross-age rg cannot be estimated from two independent samples), allowed the authors to make conclusions about similarity in the genetic architecture of diverse verbal abilities across ages, without the costly and time-consuming effort of following up the same cohort across a 9-year interval.

In research on behavioral development, twin studies have been conducted from childhood to adulthood. For example, a large twin study of parent- and self-reported fears, a common form of anxiety, was conducted on a sample assessed from age 8 years to age 20 years. They found modest evidence for genetic stability and considerable evidence for both genetic innovation and genetic attenuation from age 8 to age 20. The authors concluded that their results mostly supported ‘the‘‘developmentally dynamic’’ hypothesis [which] predicts that genetic effects on fear-proneness will vary over time’ (Kendler et al., 2008, p. 2; emphasis added).

In contrast, a longitudinal twin study of self-rated prosocial behavior in adolescence reported considerable genetic stability (albeit across a much briefer developmental gap than the above study of fears). The sample ranged in age between 13 and 17 years old (at the first assessment) and was assessed again 17 months later. The authors reported significant genetic continuity on prosocial behaviors across the two waves of data collection with a high genetic correlation (rg = 0.64) (Gregory, Light-Hausermann, Rijsdijk & Eley, 2009).

Turning to the first years of life, Gagne and Goldsmith (2011) recently assessed anger and later inhibitory control in twins at 12 and 36 months old. They reported that individual differences in anger showed considerable genetic stability from ages 12 to 36 months old (rg = 0.52), and anger at 12 months old also showed genetic overlap with later parent-rated inhibitory control at 36 months of age (rg = −0.26) (Gagne & Goldsmith, 2011). However, their findings were specific to parent report measures: they did not replicate these findings with their lab-based assessments of anger and inhibitory control. It is relevant to note that different measurement methods, such as actigraph versus parent-rated assessments of activity (Saudino, 2009) or parent- versus teacher- versus self-reports of psychopathology, for example (Ronald, Happé & Plomin, 2008; Stevenson, Asherson, Hay, Levy, Swanson, Thapar & Willcutt, 2005), have both been shown to pick up on slightly different genetic influences and, as such, measurement method always needs to be taken into account when assessing findings regarding genetic influences on development.

Longitudinal twin studies can also explore genetic stability and change on the comorbidity between two traits or disorders across development. Recent examples of this are studies on the longitudinal etiological relationship between autistic traits and anxiety-related behaviors (Hallett, Ronald, Rijsdijk & Happé, 2010) and the longitudinal etiological relationship between autistic traits, and intelligence and verbal ability (Dworzynski, Ronald, Hayiou-Thomas,, McEwen, Happé, Bolton & Plomin, 2008; Hoekstra, Happé, Baron-Cohen & Ronald, 2010). As such, these studies demonstrate to what degree the same genetic influences on one trait or disorder at an earlier age are also involved in a related trait or disorder at a later age.

In sum, evidence from developmental quantitative genetics suggests that the answer to the question in the title of this paper, ‘Is the child “father of the Man”?’, is a partial yes. The above examples demonstrate that cognitive and behavioral development show significant genetic influences that are stable across development. The answer is only a ‘partial yes’ because it is notable that all the studies discussed above also reported a degree of change in genetic influence. One limitation worthy of note is that longitudinal twin studies that span both childhood and adulthood are relatively rare due to the extensive research effort and time that they inevitably require (several decades of research and funding).

Neurodevelopmental genetic syndromes

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

The study of neurodevelopmental genetic syndromes with a known genetic cause has been realized to be a powerful tool for understanding how genetic influences operate across development (e.g. Paterson, Brown, Gsodl, Johnson & Karmiloff-Smith, 1999) (in addition, of course, to the important goal of understanding these conditions in their own right). Because the genetic cause of the condition is known, the developmental phenotype can be associated with a specific change in the genome. Furthermore, because individuals with these conditions can be diagnosed and studied prospectively from infancy onwards, this provides greater scope to study development than in multifactorial conditions where diagnosis only occurs at a later age, and as such the infancy/early childhood developmental window is often missed or poorly understood. Developmentalists have placed emphasis on the need to understand genetic effects on the developing infant brain, which is very different from the adult brain in terms of degree of specialization and localization (Karmiloff-Smith, 2007).

Williams syndrome will be discussed here as an example of how a known genetic developmental syndrome can inform us about genetic influences across development. Williams syndrome involves a known genetic microdeletion on the long arm of chromosome 7 affecting as many as 28 genes and occurs in 1 in 15,000 live births. The Williams syndrome phenotype involves a specific physical, behavioral and cognitive profile (Donnai & Karmiloff-Smith, 2000). Behavioral features include a strong preference for faces and inappropriate friendliness to strangers. Notable cognitive features include an IQ of approximately 60 with higher verbal compared to visuo-spatial abilities and serious numerical difficulties. Individuals with Williams syndrome have some atypical visuo-spatial abilities such as difficulties with visual search in toddlerhood (Scerif, Cornish, Wilding, Driver & Karmiloff-Smith, 2004), impairments with identifying spatial relationships between landmarks in older childhood (Farran, Blades, Boucher & Tranter, 2010) and difficulties with multiple object tracking in childhood and adulthood (O’Hearn, Hoffman & Landau, 2010).

An example of research on Williams syndrome that has considered how the phenotype varies across the whole of development is work on number processing. Studies have demonstrated that individuals with Williams syndrome show atypical large number processing both in infancy, as measured using an experimental observational task to test small versus large number discrimination (Van Herwegen, Ansari, Xu & Karmiloff-Smith, 2008), and numerical impairments in later childhood and early adulthood as assessed using arithmetic assessments (Udwin, Davies & Howlin, 1996).

A second example of how the Williams syndrome phenotype has been investigated across development is a study on the personality and behavior profile of Williams syndrome individuals (Gosch & Pankau, 1997). Individuals with Williams syndrome have a distinctive profile of personality characteristics. Gosch and Pankau’s (1997) study included three groups of individuals with Williams syndrome: under 10-year-olds, 10–20-year-olds, and over 20-year-olds. While some personality measures, such as over-friendliness, showed a lower mean score in the adolescent and adult Williams syndrome groups compared to the under 10-year-old group (that is, scores became closer to general population averages), scores were still higher than population means. Overall the authors concluded that the characteristic personality profile of individuals with Williams syndrome was stable from childhood to adulthood.

In answer to the question in the title of this paper, ‘Is the child “father of the Man”?’, findings from research on Williams syndrome suggest that the answer is yes to the extent that there appears to be relative consistency in the phenotype profile across the lifespan in this neurodevelopmental syndrome with a known genetic cause. Although known genetic syndromes are a different genetic 'model' to that of multiple probabilistically acting genes on common traits, they are informative about the effects of variation in the genome on development. However, there is the consideration of to what degree findings from specific neurodevelopmental genetic syndromes with a known genetic cause generalize to understanding the genetic influences on cognitive and behavioral traits in the general population. On a practical level, samples tend to be small and heterogeneous, they only directly inform us about the specific area of the genome in which the mutation has occurred, and developmental fluctuation in environmental influences is not usually included within the study designs, although cross-cultural studies have been conducted (see Zitzer-Comfort, Doyle, Masataka, Korenberg & Bellugi, 2007, for an example). In the next section, candidate gene studies of infant and adult samples demonstrate how typically developing samples can be used to test the degree of genetic continuity across the lifespan.

Developmental candidate gene association studies

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

Candidate gene studies involve testing for an association between a trait or disease of interest and a known candidate gene. The known candidate gene might be selected because it is ‘biologically plausible’ in that it codes for a protein that is hypothesized to have a role in the phenotype’s causal pathway, or selected on the basis of genomic position (e.g. from linkage studies).

Candidate genes that play a role in the dopamine and serotonin neurotransmitter pathways have been selected as biologically plausible for many cognitive and behavioral traits (Hirschhorn, Lohmueller, Byrne & Hirschhorn, 2002). These candidate gene studies can inform us about genetic influences on development because associations between phenotypic variation and specific genetic variation can be explored at different ages. To illustrate this, some recent studies are highlighted that have tested whether genes that play a role in the dopamine and serotonin neurotransmitter pathways are associated with similar behavioral phenotypes in infancy as well as later life.

Genetic variation in a region of a gene that encodes tryptophan hydroxylase isoform 2 (TPH2) is thought to modify gene expression in such a way as to alter serotonin concentrations in neurons in the brain. In adults, variation in this gene region has been associated with attention regulation and cognitive control (e.g. Strobel, Dreisbach, Muller, Goschke, Brocke & Lesch, 2007). In a recent study, the same genetic variation (a greater number of T alleles) that is associated with poorer performance on attentional tasks in adults was associated with a higher number of missing attention shifts in 7-month-old infants in an experimental paradigm that tested for efficiency of attentional shifts (Leppanen, Peltola, Puura, Mantymaa, Mononen & Lehtimaki, 2011). These studies appear to suggest that the same genetic variation influences attentional processes in infancy and adulthood.

The monoamine oxidase A (MAOA) gene codes for a catabolic enzyme that is involved in regulating the degradation of serotonin and dopamine and as such is considered another biologically plausible candidate gene for cognitive and behavioral traits. Variation in the MAOA gene has been associated with impulsivity and aggressive behavior in adults (e.g. Brunner, Nelen, Breakefield, Ropers & van Oost, 1993). In a recent study, 6-month-old infants were assessed for their self-regulation in the lab by observing the degree to which they oriented away from a threatening event (the presentation of a large toy chimpanzee). The authors concluded from this first study of MAOA in infancy that a common functional MAOA variable number tandem repeat (MAOA-uVNTR) was associated with variation in self-regulatory behavior (the more active genotype, 4/4, was associated with higher regulation), although the association was only found in girls and not boys (Zhang, Chen, Way, Yoshikawa, Deng, Ke, Yu, Chen, He, Chi & Lu, 2011).

A third example is a recent study of dopamine system genes in infancy. The catechol-O-methyltransferase (COMT) gene codes for an enzyme which is involved in the metabolic degradation of dopamine. The COMT Val158 Met polymorphism has been associated with efficiency of function in prefrontal cortex in school-age children (Diamond, Briand, Fossella & Gehlbach, 2004) and adults (Mier, Kirsch & Meyer-Lindenberg, 2009). One of the key findings in a recent study of infants was that variation in the COMT Val158 Met polymorphism was associated with degree of distractibility on the ‘Freeze-Frame’ task, a task specifically designed to assess attention and frontal cortex functioning, in 9-month-old infants (Holmboe, Nemoda, Fearon, Csibra, Sasvari-Szekely & Johnson, 2010) (although variants in other candidate genes, such as the DRD4 48-bp VNTR, did not show an association with performance on this task). Thus several recent studies suggest that there is some continuity in the genotype–phenotype associations found in infancy, childhood and adulthood.

Finally, two studies from a Dutch longitudinal sample assessed from childhood through to adolescence tested whether several serotonin and dopamine system-related genes, including TPH2 and genes encoding serotonin transporters, were associated with longitudinal measures of attention problems, and anxiety and depression, respectively (Middeldorp, Slof-Op’t Landt, Medland, van Beijsterveldt, Bartels, Willemsen, Hottenga, de Geus, Suchiman, Dolan, Neale, Slagboom & Boomsma, 2010; van Beijsterveldt, Middeldorp, Slof-Op’t Landt, Bartels, Hottenga, Suchiman, Slagboom & Boomsma, 2011). Perhaps surprisingly, however, these longitudinal behavioral traits did not show an association with the selected candidate genes.

The answer to the question ‘is the child “father of the Man”?’ appears to be yes to some extent from the example candidate gene association findings on dopamine and serotonin system genes mentioned above. There is some evidence for the same genotype–phenotype associations emerging from infancy, childhood and adulthood. These candidate gene association studies are perhaps the most direct, if restricted, examples of genetic stability across development because they take one genetic variant and one phenotype and test for their association at two age ranges, e.g. in infancy and adulthood. They are by default limited to the phenotype, the age groups, and the restricted amount of genetic variation on which they focus. These studies ignore the rest of the genome, environmental variation and other phenotypes. It has been argued, in the field of attentional processes for example, that research needs to be carried out on ‘gene × environment × time interactions across domains’ (Scerif, 2010, p. 811, emphasis added; see also Posner, Rothbart & Sheese, 2007). This is an admirable albeit ambitious goal. Candidate gene association studies offer a starting place for investigating main effects on biologically plausible genetic associations.

The most pertinent limitation to candidate gene studies is that the significant findings have proved difficult to replicate. The reason for this could be that the gene effect sizes were smaller than predicted and so studies were underpowered to replicate them, but there are other possible reasons (Hirschhorn et al., 2002; Plomin, in press). As reviewed in the next sections, GWAS and DNA resequencing offer more comprehensive and systematic methods for finding replicable genetic associations across development.

Developmental genome-wide association studies

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

‘Genome-wide’ association studies (GWAS) involve simultaneously testing for associations between hundreds of thousands of common variants, that is, variants with minor allele frequencies greater than 1%, spread across the entire genome (in contrast to looking at a single ‘candidate’ gene) with a trait or disorder. This design has been made possible by the arrival of microarrays (Plomin & Schalkwyk, 2007). GWAS are more systematic than candidate gene studies because if a microarray with good coverage of the genome is used, most of the common variation in the genome will be measured.

GWAS of individual phenotypes have struggled to find genetic variants associated with cognitive and behavioral phenotypes, a phenomenon named the ‘missing heritability’ problem in the GWAS field at large (Manolio, Collins, Cox, Goldstein, Hindorff, Hunter et al., 2009). An exemplar comes from GWAS of ADHD. ADHD is one of the most highly heritable behaviorally defined disorders, yet a recent and well-designed study, and further meta-analysis, reported no genome-wide significant genetic associations (Neale, Medland, Ripke, Anney, Asherson, Buitelaar et al., 2010a; Neale, Medland, Ripke, Asherson, Franke, Lesch et al., 2010b).

Could the success of GWAS be improved by incorporating findings from quantitative genetics? An example of how findings from quantitative genetics have fed into the design of GWAS comes from research on autistic traits. Twin studies have shown that social autistic traits and restricted repetitive behaviors and interests (RRBIs, also called non-social autistic traits) are both types of autistic symptoms that show high heritability but appear to have largely distinct genetic and environmental influences in middle childhood (Happé & Ronald, 2008; Ronald, Happé, Bolton, Butcher, Price, Wheelwright, Baron-Cohen & Plomin, 2006; Ronald, Happé & Plomin, 2005). On the basis of this finding of a lack of genetic overlap between social and RRBI autistic traits, the first GWA study of autistic traits was conducted separately for social and non-social autistic traits (Ronald, Butcher, Docherty, Davis, Schalkwyk, Craig & Plomin, 2010).

However, quantitative genetic research findings regarding genetic stability and change do not appear to have fed into GWA study designs. At the time of writing, and based on searches in Pubmed and http://www.genome.gov/gwastudies/, there appeared to be no GWAS on cognitive or behavioral development that employed a longitudinal repeated measures phenotype. This is despite many developmental twin studies reporting evidence for some genetic change across development, as demonstrated in the Developmental quantitative genetics section above. A prescient study 6 years ago took five single nucleotide polymorphisms (SNPs) that had been identified as associated with cognitive ability at age 7 and combined them into an SNP set composite to test for association with cognitive ability in earlier childhood (Harlaar, Butcher, Meaburn, Sham, Craig & Plomin, 2005). However, to the author’s knowledge, no full-scale GWAS of cognitive or behavioral traits have tested such developmental hypotheses.

In research on health-related phenotypes, GWAS have begun to incorporate longitudinal data; for example, there has been a longitudinal GWA study of cardiovascular disease risk factors (Smith, Chen, Kahonen, Kettunen, Lehtimaki, Peltonen et al., 2010). The authors’ conclusions are also relevant to the field of cognitive and behavioral traits: ‘longitudinal studies may be a useful tool to better capture time-dependent variation that could ultimately be [more] predictive of future outcomes’ (Smith et al., 2010, p. 6; emphasis added).

Is the reason for the lack of significant associations in many GWAS of cognitive or behavioral traits due to their reliance on cross-sectional data that do not capture all the ‘time-dependent’ genetic variation? This remains a possibility, although clearly it is unlikely to be the only reason (see Manolio et al., 2009, and Plomin, in press, for a discussion of other potential reasons for the lack of significant findings). With regard to the ADHD GWAS mentioned above (Neale et al., 2010a; Neale et al., 2010b), the ADHD diagnosis is thought to be moderately stable. A developmental quantitative genetic twin study reported genetic correlations between ADHD traits of between 0.32 and 0.62 from age 2 to age 8. These results on ADHD traits suggest considerable genetic stability but also some genetic change across childhood (Kuntsi, Rijsdijk, Ronald, Asherson & Plomin, 2005) which could be taken into account in future GWAS of ADHD traits. Future GWAS of cognitive or behavioral traits and disorders could benefit from assessing whether the trait being studied shows genetic change across ages, as Smith et al. described, ‘to better capture time-dependent [genetic] variation’ (2010, p. 6; emphasis added).

In answer to the question ‘Is the child “father of the Man”?’, developmental GWAS of cognitive and behavioral traits have not yet attempted to answer this. The GWAS field is struggling with the problem of missing heritability and the focus at present is on obtaining larger samples to detect small genetic effects (Manolio et al., 2009). Here we highlight another consideration for the design of GWAS, which is the potential value of longitudinal data for phenotypes that are influenced by partly different sets of genes across development. Capturing more of the genetic variation acting across development might not help with the small effect size issue but it would help to provide a more accurate measure of the genetic variation that influences a trait across ages. A suggested rule of thumb for designing GWAS that will capture maximum cross-age genetic variation could be that if the genetic correlation between any two ages is greater than 0.5, researchers can justify using data from a single age because they know that a considerable amount of genetic variation is the same across the two ages. If the genetic correlation is less than 0.5, researchers should endeavor to capture the genetic variation present at both ages in their design, for example, by conducting a GWA study separately for each age, or by using structural equation models to incorporate the longitudinal data in the analysis. In time, it is predicted that the value of formally modeling development in GWAS will be recognized (e.g. Das, Li, Wang, Tong, Fu, Li, Xu, Ahn, Mauger, Li & Wu, 2011).

DNA resequencing

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

DNA resequencing involves identifying the entire sequence of DNA code in an individual’s genome. The cost of DNA resequencing has fallen dramatically in recent years and the technical ability has improved, which allows resequencing of vast quantities of DNA to be carried out accurately and in great depth. These developments have led to speculation that it will not be long before many individuals’ DNA sequence data are collected routinely (Collins, 2010). Plomin (in press) predicted that there will be two ways in which this sequencing revolution will affect developmentalists. First, it will democratize the whole genome so that it is not just genes themselves that are the focus (coding genes make up less than 2% of the genome, and as reviewed above, candidate genes narrow the focus even further to just a handful of genes). DNA sequence data provide information about the entire genome, which includes rare variants, mutations, and non-coding genes. Second, when sequence data do become available there will be no need for DNA to be collected more than once or for genotyping to be conducted. This should revolutionize the field of genetic influences on development and allow us to answer the question of this paper, ‘Is the child “father of the Man”?’ in terms of genetic influences with much greater certainty.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

This review paper addressed the question posed at the start of this paper, namely in terms of genetic influences, ‘Is the child “father of the Man”?’ or phrased differently, to what extent do the same genes influence individual differences on the same traits in infancy, childhood, adolescence, and adulthood? Evidence from developmental quantitative genetic and developmental candidate gene studies both support the notion that to some extent the same genes influence early and later cognitive and behavioral traits (from infancy to adulthood), although we note that developmental quantitative genetic studies also report some evidence for changes in genetic influence and that not all candidate gene associations replicate across ages. Research on known genetic syndromes such as Williams syndrome shows that individuals with a known genetic mutation display a phenotype associated with this mutation that is relatively stable across life from infancy to adulthood. It is now time for the newer, more technology-enabled fields such as GWAS and DNA resequencing to take on board the dynamic nature of human behavior and thereby further our understanding of the degree of genetic stability across development.

Key points

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References
  • • 
    Genetic influences can be stable across ages (‘genetic continuity’) or new genetic influences can occur at one age and not another (‘genetic change’, which can involve either genetic innovation or genetic attenuation).
  • • 
    Longitudinal twin studies of cognitive and behavioral traits tend to report both genetic stability and change across development, although twin studies spanning the entirety of human development (infancy to adulthood) are rare.
  • • 
    Candidate gene association studies offer a starting place for investigating main effects on biologically plausible genetic associations across development (for example, dopamine and serotonin system genes have been associated with apparently similar infant and adult phenotypes), but newer technologies (genome-wide association studies and DNA resequencing) now provide a far more systematic approach to explore genetic associations.
  • • 
    Genetic syndromes with known etiology can be studied from infancy to adulthood, although the informativeness of these studies to genetics is to some extent limited to the region of the genetic mutation. Research on Williams syndrome is an example that supports the argument for relative genetic stability across development and is a good example of the leverage offered by known genetic syndromes for studying genotype–phenotype mapping across development.
  • • 
    No genome-wide association study on cognitive or behavioral traits has yet employed longitudinal phenotypic data.
  • • 
    Knowledge of how genes operate across development will have important practical consequences in informing, for example, the development of preventative strategies for heritable conditions, as well as impacting basic science by contributing to our understanding of the causes of individual differences in behavior.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References

Thanks to Miss Sarah Hawley, Professor Mark Johnson, Professor Annette Karmiloff-Smith, Dr Emma Meaburn, Professor Robert Plomin, Dr Elise Robinson, Miss Aline Scherff and Mr Mark Taylor for their helpful comments on earlier versions of this manuscript.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Developmental quantitative genetics
  5. Neurodevelopmental genetic syndromes
  6. Developmental candidate gene association studies
  7. Developmental genome-wide association studies
  8. DNA resequencing
  9. Conclusions
  10. Key points
  11. Acknowledgements
  12. References