(This article is best read after listening to "How genes make us what we are")
Comments on new babies such “He’s got his mother’s eyes” reflect our intuitive understanding that many human characteristics are inherited from our parents and therefore under genetic control. These include physical attributes such as height or eye colour, behavioural traits such as IQ or anxiety, and clinical diseases such as asthma, heart disease or cancer. The particular form any trait takes in an individual is referred to as their phenotype in genetic studies.
Recent developments in genetic technology have brought us closer to identifying many specific genes and the phenotypes they influence. Because of variations in genetic structure in the population, different individuals will have slightly different versions of any given gene. The specific genetic variant possessed by an individual is known as their genotype. For many genes, we know that differences between individuals in their genotypes contribute to differences between them in their phenotypes.
In conducting our study, we have used genetic markers. A marker is usually a DNA sequence that:
(1) is polymorphic (i.e known to vary between individuals)
(2) occurs at a known position on a known chromosome
In our DNA samples from singers and non-musicians, we tested a number of markers in the laboratory. We were interested in detecting genotypes that are more common in choral singers than in non-musicians. Any genotypes more common in singers might be so because they influence musical ability. However "being in a choir", and all the other phenotypes listed at the start of this article, are termed ‘complex’. This means that they are controlled by genetic variation at many different sites in the human genome. No person is blue eyed, asthmatic or a choral singer because of their genotype in respect of a single point ("locus") on the genome.
How are these loci identified? Genetic association studies like ours usually ascertain and study the DNA of two groups of people – those with, and those without, the trait of interest. This method is called a ‘case control’ study. Any genotype that differs in frequency between the two groups may affect the presence or absence of the trait.
In the example on the right (Table 1), we have used a theoretical single nucleotide polymorphism, or SNP, as a genetic marker. An SNP comprises a single DNA base and, in this example, the SNP exists as a C in some individuals or a T in others. Each individual carries two versions, or alleles, at this SNP - one inherited from each parent. So, each individual can carry the combination of two C alleles (genotype CC), two T alleles (genotype TT) or one copy of each allele (genotype CT).
Let us say we test this theoretical SNP in our own sample of 500 individuals, where the trait of interest is “being in a choir”. We can use the table to display the numbers of each genotype for those with the trait (‘cases’, or choral singers) and those without (‘controls’, or non-musicians) . The lower case letters in the table indicate the numbers observed ( a = number of cases carrying CC genotype).
For a straightforward statistical analysis, the data are often summarised as the number of copies of the C and T allele observed in each group (Table 2).
The numbers of C alleles in the cases and the controls will differ slightly. Is the difference between (2a+b) and (2d+e) small enough to be due to random chance - similar to the variation seen in tossing a coin? Or is the difference large enough to indicate a systematic difference between case and controls, suggesting that this genotype is correlated with choir membership? Answering this question is the crux of determining whether the genetic marker tested is associated with the trait.
A statistical test is performed to test whether the numbers of C alleles observed in cases and controls in our sample is consistent with random chance. The test produces a p-value. This value is the probability of these numbers of C and T alleles (or more extreme numbers) occurring in a sample of this size, if the proportion of C alleles in cases and controls is actually equal in the whole population. A p-value of 0.05 is often used a threshold for statistical significance. A p-value greater than 0.05 indicates the result seen in the sample is consistent with random chance, a p-value less than 0.05 suggests that genotype at this marker does indeed affect the trait. This traditional threshold reflects a 1 in 20 chance that the sample result was obtained when there is, in fact, no difference in allele frequencies between cases and controls in the population.
Replication is the corner-stone of genetic studies. Any association needs to be found in another group of individuals in whom the trait has been defined in exactly the same way, with their DNA genotyped in a different laboratory. Many of the markers tested in the choral singing study have previously been associated with other forms of performance or with musical aptitude. Further studies – both association studies like ours, and functional studies looking at the actual properties of given genotypes– may be necessary before their effect on choral singing can be taken as proven.
Any genotypes statistically associated with being a choral singer in our study will be part of a spectrum of genetic, and perhaps non-genetic properties, that contribute to this phenotype. The number of markers required to determine with high accuracy an individual’s status as case or control depends on the trait. For eye colour, just a few markers can be used to discriminate between blue and brown eye colour [1 , 2]. For Crohn’s disease (a serious gastroenterological disease), variants in 71 genes have been identified, but these account for only a small proportion of the genetic component . As stated above, choral singing is likely to involve many genes, and extensive studies will be needed to identify them.
The study of the genetic contribution to human characteristics is only just beginning, and future studies will identify intriguing insights into how our genetic inheritance shapes our appearance, our behaviour, our medical history, or, in this case, our inclination to choral singing. Our understanding of genetics has moved on from the concept of ‘a gene for’ musicality, height, or asthma, and we now know that each of these traits has a complex pattern of susceptibility involving many genetic variants. We have the genetic knowledge, the molecular technology and the statistical tools to identify these variants and the next few years should see these uncovered at a rapid pace.
King’s College London
 Liu, F., Van Duijn, K., Vingerling, J.R., Hofman, A., Uitterlinden, A.G., Janssens, A.C., and Kayser, M. (2009). Eye color and the prediction of complex phenotypes from genotypes. Curr. Biol. 19, R192–R193.
 Sturm RA, Larsson M. Genetics of human iris colour and patterns. Pigment Cell Melanoma Res. 2009 Oct;22(5):544-62.
 Franke A, McGovern DP, Barrett JC, Wang K, et al. Genome-wide meta-analysis increases to 71 the number of confirmed Crohn's disease susceptibility loci. Nat Genet. 2010 Dec;42(12):1118-25.