Search tips
Search criteria 


Logo of plosonePLoS OneView this ArticleSubmit to PLoSGet E-mail AlertsContact UsPublic Library of Science (PLoS)
PLoS One. 2012; 7(12): e37385.
Published online 2012 December 26. doi:  10.1371/journal.pone.0037385
PMCID: PMC3530597

Genetic Copy Number Variation and General Cognitive Ability

Courtney G. Montgomery, Editor


Differences in genomic structure between individuals are ubiquitous features of human genetic variation. Specific copy number variants (CNVs) have been associated with susceptibility to numerous complex psychiatric disorders, including attention-deficit-hyperactivity disorder, autism-spectrum disorders and schizophrenia. These disorders often display co-morbidity with low intelligence. Rare chromosomal deletions and duplications are associated with these disorders, so it has been suggested that these deletions or duplications may be associated with differences in intelligence. Here we investigate associations between large (≥500kb), rare (<1% population frequency) CNVs and both fluid and crystallized intelligence in community-dwelling older people. We observe no significant associations between intelligence and total CNV load. Examining individual CNV regions previously implicated in neuropsychological disorders, we find suggestive evidence that CNV regions around SHANK3 are associated with fluid intelligence as derived from a battery of cognitive tests. This is the first study to examine the effects of rare CNVs as called by multiple algorithms on cognition in a large non-clinical sample, and finds no effects of such variants on general cognitive ability.


Among humans, individual differences in measured intelligence are associated with important life outcomes, including long-term health and wellbeing [1], [2]. General cognitive ability (usually denoted g) is a quantitative trait, and is assessed using cognitive ability tests. Empirical evidence for g was first described by Spearman [3] who found that diverse mental capabilities tended to show positive covariation. The general intelligence factor, g, accounts for about 50% of the total variance when a number of diverse mental tests are administered to population samples [4]. The g factors derived from very different cognitive test batteries rank people almost identically [5], and intelligence differences are highly stable across the human lifecourse [6]. Two major facets of intelligence are crystallized (gc) and fluid (gf) intelligence. Crystallized-type intelligence is characterised by a relative lack of decline with age [7], and is typically assessed using tests of acquired knowledge, most often vocabulary. Types of cognitive ability that are termed ‘fluid intelligence’ tend to decline with age from young or middle adulthood [7], and are assessed using unfamiliar, sometimes abstract materials, and involve on-the-spot thinking, often under time pressure, and rely relatively little on prior knowledge. Intelligence is substantially heritable [8], [9] and, although variants in a number of candidate genes have shown significant associations, few have replicated [10], [11]. This is true for many complex traits: even in the most highly heritable, such as height, known variants account for only a small proportion of the observed heritability [12]. Several hypotheses have been proposed as to where this “missing heritability” resides [13], [14]. These include common variants of small effect [15], rare variants with large effects [16], epistatic interactions [17], epigenetic factors [18] as well as other forms of genetic variation beyond single nucleotide polymorphisms (SNPs). One example of this last factor is structural genetic variation, which includes copy number variation.

Copy number variants (CNVs) are defined as segments of DNA longer than 1kb present in variable numbers of copies across individuals within a population sample [19]. CNVs are defined relative to a normal copy number of two; hence, segments that are present in more than two copies within an individual are classed as duplications, and fewer than two are classed as deletions. CNVs are observed ubiquitously throughout the genomes of humans [20] and other organisms [21]. Older SNP genotyping arrays missed some of this variation, but recent high density arrays include multiple non-polymorphic markers in regions of known structural variation, allowing more reliable detection of CNVs.

The Database of Genomic Variants [22] reports over 60,000 CNVs at more than 15,000 loci from 42 reported studies, that collectively cover more than a third of the human genome. Whereas 1kb is typically taken as the minimum length for a CNV, the largest can span several megabases, and can potentially disrupt multiple genes and/or regulatory regions, each of which may have an effect on gene expression and phenotype [23]. Initial studies of CNV prevalence [20] suggested extensive numbers of smaller CNVs across populations. There is still some debate regarding the best method to detect CNVs from SNP data [24], [25]. Different methods show marked variation in the number and extent of CNVs detected from the same samples, and the reliability of calling shorter CNVs is especially questionable. However, calls for longer CNVs are more consistent between methods [26], are more likely to represent true variants, and thus have the potential for more robust replication. Therefore, the current study examines the effect of longer, rare CNVs on human intelligence differences.

Specific CNVs have been associated with susceptibility to illnesses including HIV-1 infection [27], autoimmune disorders [28][31], and cancer [32]; nervous system disorders [33] such as Alzheimer's Disease [34], Parkinson's Disease [35], [36], epilepsy [37]; and psychiatric disorders, including schizophrenia [38], [39], mental retardation [40][42], autism [43][46] and major depressive disorder [47]. However, as most detected CNVs are relatively rare within discovery cohorts [48], association tests for individual CNVs will be of limited power to detect significant variants. An alternative approach is to test jointly for the effect of multiple rare CNVs on disease status, by comparing the overall CNV load between cases and controls. The overall CNV load can be measured as the total number of rare CNVs carried, the total length of these CNVs, or the total number of genes they disrupt. The method of examining overall CNV load has been applied to a number of psychiatric and neurological disorders [49][51]. Here, we apply it to variation in human intelligence, treated as a quantitative trait, measured in community-dwelling older people.

A report by Yeo et al. [52] identified a significant association between the extent of rare genetic deletions and Full-Scale Intelligence Quotient (FSIQ), derived from the Wechsler abbreviated scale of intelligence [53], in a very small clinical sample of patients undergoing treatment for alcoholism. The authors acknowledged that there have also been findings of CNV differences between controls and disorders that involve cognitive deficits. Williams et al. [54] examined the effect of rare CNVs on risk of Attention-Deficit Hyperactivity Disorder (ADHD), finding significant differences in numbers of CNVs ≥500 kb between cases and controls for total CNV burden. Patients with psychiatric disorders such as schizophrenia and ADHD often display cognitive deficits [55], suggesting that the burden associated with rare CNVs may also have an effect on intelligence itself. Here we present analyses of the effect of large, rare CNVs on measured intelligence in cohorts of relatively healthy individuals with a total sample size of more than 3,000 older people.


Ethics Statement

Ethical approval for the Lothian Birth Cohort 1921 study was obtained from the Lothian Research Ethics Committee. Ethical approval for the Lothian Birth Cohort 1936 study was obtained from Scotland's Multicentre Research Ethics Committee and the Lothian Research Ethics Committee. Ethical approval for the Aberdeen Birth Cohort 1936 study was obtained from the Grampian Research Ethics Committee. Ethical approval for Manchester and Newcastle Longitudinal Studies of Cognitive Ageing Cohorts study was obtained from the University of Manchester. Written consent was received from all participants for their information to be stored in the relevant university database and used for research.

Cohort Descriptions

Lothian Birth Cohort 1921

The Lothian Birth Cohort 1921 (LBC1921) is a longitudinal study of cognitive ageing conducted at the University of Edinburgh. Individuals in the LBC1921 were born in 1921 and were recruited and tested in old age, as described elsewhere [6], [56]. In total, 550 individuals (234 male, 316 female) were tested at mean age 79.1 years (SD = 0.6). Participants were tested individually, and completed a battery of cognitive tests including: The Moray House Test No. 12 (MHT) [57], Raven's Standard Progressive Matrices [58], Verbal Fluency [59], and Logical Memory [60]. Participants also completed the National Adult Reading Test (NART) [61].

Lothian Birth Cohort 1936

The Lothian Birth Cohort 1936 (LBC1936) is a longitudinal study of cognitive ageing conducted at the University of Edinburgh. Individuals in the LBC1936 were born in 1936 and were recruited and tested in old age, as described elsewhere [62]. In total, 1091 individuals (548 male, 543 female) were tested at a mean age of 69.5 years (SD  = 0.8). Participants were tested individually on a large battery of cognitive tests [62] including the MHT, and the following six tests from the Wechsler Adult Intelligence Scale-IIIUK (WAIS-IIIUK [60]): Digit Symbol Coding, Block Design, Matrix Reasoning, Digit Span Backwards, Symbol Search, and Letter-number Sequencing. Participants also completed the NART [61].

Aberdeen Birth Cohort 1936

The Aberdeen Birth Cohort 1936 (ABC1936) is a longitudinal study of cognitive ageing. Individuals in the ABC1936 were born in 1936 and were recruited and tested in old age as described elsewhere [6], [56]. In total, 498 individuals (243 men, 255 women) were tested at mean age 64.6 years (SD = 0.9). Cognitive tests completed were the NART [61], Raven's Standard Progressive Matrices [58], Rey Auditory Verbal Learning Test (AVLT) [59], Digit Symbol and Block Design sub-tests of the Wechsler Adult Intelligence Scale-Revised [63], and the Uses of Common Objects Test [59].

Manchester and Newcastle Longitudinal Studies of Cognitive Ageing Cohorts

The University of Manchester Age and Cognitive Performance Research Centre (ACPRC) programme began in 1983 and has documented longitudinal trajectories in cognitive function in 6371 older adults in the North of England. The group comprises 1917 men and 4454 women with mean age 65.6 years (SD = 14.3) at initial recruitment. Details of the battery of cognitive function tests used in alternating batteries can be found in Rabbitt et al. [64]. The Dyne Steel DNA Archive for Ageing and Cognition was established following invitation to all participating volunteers 1999 and 2004. This resulted in 1829 volunteers attending Manchester or Newcastle Universities, or being visited at home for blood sample collection.

Construction of Cognitive Phenotypes

We constructed cognitive phenotypes of fluid- and crystallized-type intelligence for each of the cohorts. To represent crystallized intelligence (gc), we used the NART in the Lothian Birth Cohorts of 1921 and 1936, and the Aberdeen Birth Cohort of 1936, and the Mill Hill Vocabulary A and B vocabulary tests in the Manchester and Newcastle cohorts. All are vocabulary-based tests and are good representatives of the underlying construct of crystallized intelligence. The fact that not all cohorts received precisely the same vocabulary test introduces a phenotypic heterogeneity that is only likely to slightly reduce the size of any observed association between CNV indices and intelligence.

A general intelligence factor for fluid-type intelligence was derived in the Scottish cohorts using principal components analyses (PCA), with higher values of the components reflecting better ability. Strictly speaking, PCA does not produce ‘factors’, but this is a common usage. For the two Lothian Birth Cohorts, and the Aberdeen Birth Cohort of 1936, the scores on a number of fluid-type intelligence tests were subjected to PCA. In all cases, inspection of the scree slope and the Eigenvalues-greater-than-one criterion indicated a single component that was then extracted. Individuals' scores on the first unrotated principal component were used to represent fluid-type general intelligence (gf). In LBC1921, the tests used to construct gf were the Moray House Test, Raven's Matrices, Logical Memory, and Verbal Fluency. In LBC1936, the six tests from the WAIS-IIIUK described above were used to construct gf. The tests used to define gf in ABC1936 were Raven's Progressive Matrices, Digit Symbol, Uses of Common Objects, and AVLT. The first principal component accounted for 49% of variance in ABC1936, 56% in LBC1921 and 51% in LBC1936. The range of tests administered to the LBC1936 sample allowed the construction of the same gf battery used in LBC1921 using the LBC1936 data, and the correlation between the gf scores derived from two different sets of tests on LBC1936 was ~0.7. For the Manchester and Newcastle cohorts, a general fluid-type intelligence ability factor, gf, was obtained from a random effects model fitted by maximum likelihood to the standardized age regressed residuals obtained for each sex from tests including the Alice Heim 4 (AH4) parts 1 and 2 general intelligence tests. Detailed task descriptions can be found in Rabbit et al. [63]. The gf scores were derived separately for males and females in the Manchester and Newcastle cohorts. Although different tests were used to construct the general fluid intelligence factors between cohorts, the correlation between such factors when they are derived on the same sample tends to be very high [5], [65]. All phenotypes described above were corrected for age, and for sex for those not derived separately by gender. Standardized residuals were used in all subsequent analyses.

SNP Genotyping and Quality Control

Genomic DNA was isolated using standard procedures at the Wellcome Trust Clinical Research Facility (WTCRF) Genetics Core, Western General Hospital, Edinburgh for LBC1936 and ABC1936, and Medical Research Council Technology, Western General Hospital, Edinburgh for LBC1921. The UK DNA Banking Network was used for the Manchester and Newcastle Dyne-Steele samples. In total, 3782 samples were genotyped at the WTCRF Genetics Core ( using the Illumina610-Quadv1 chip (LBC1936 N = 1,042; LBC1921 N = 526; ABC1936 N = 456; Manchester N = 901; and Newcastle N = 877).

Samples were subjected to the following quality control (QC) procedures: individuals where self-reported gender disagreed with genetic evidence were removed. Pairwise IBD between individuals was estimated and, where it was greater than 0.25, one of each pair was removed from the analysis. Samples with SNP call rate <0.95, and those showing evidence of non-Caucasian origin by multi-dimensional scaling were also removed. After QC, a total of 3,511 samples remained (LBC1936 N = 1,005; LBC1921 N = 517; ABC1936 N = 426; Manchester N = 805; and Newcastle N = 758). SNPs were retained for analyses that met the following criteria: call rate ≥0.98, minor allele frequency ≥0.01, and Hardy-Weinberg Equilibrium test with P≥0.001. A total of 549,754 SNPs passed these QC criteria and, with the inclusion of a further 21,890 non-polymorphic markers in known CNV regions, 571,644 markers in total were used to call CNVs.

CNV Calling and Quality Control

CNV calling used Log-R-Ratio (LRR) and B-allele Frequency (BAF) values normalised and extracted from raw signal data using Illumina's Genome Studio software. CNV calling was performed using the script from PennCNV [66], and the QuantiSNP package [67]. Only variants that were called by both algorithms were used in the analysis. Where CNV boundaries were not identical between the two algorithms, the start and end of the overlapping region were taken as the CNV boundaries. Quality control steps [54] were applied at the level of sample quality, and of individual CNVs. Twenty samples were genotyped in duplicate to check consistency of genotype calling. Before QC steps were implemented, 47% of total CNVs called were common to both samples, increasing to 67% as restrictions on CNV length (≥ 500kb) were put in place. After QC, the sample in each pair with the lower SNP call rate was excluded from further analysis. To investigate the effect of rare CNVs, the QC and selection procedures of Williams et al. [54] were followed. Briefly, samples with standard deviation of Log-R Ratio across all markers greater than 0.3 were excluded, as were samples with 30 or more CNVs called at longer than 100kb, due to the unreliability of these calls. Quality control was also performed at the level of individual CNVs, with variants that spanned fewer than 15 contiguous markers discarded. Any adjacent CNVs that appeared to be artificially separated by the calling algorithm were merged: CNVs were candidates for merging when pairs of adjacent variants on the same chromosome, of the same copy number state, were greater than 200kb in length and separated by a distance of less than half the total length of the merged variant. LRR and BAF data for all candidate merges were inspected visually. To investigate the effect of only rare variants, CNVs present in greater than 1% of each cohort were discarded from analysis. CNV boundaries do not necessarily correspond exactly between samples, so variants were removed where any marker along their length was called in a CNV region in greater than 1% of each cohort. We investigated the effect of removing common CNVs using the 1% criterion on the entire sample, rather than individual cohorts, and found no differences in the significance of results between the two sets of data (not shown).

Modelling CNV load

To investigate the effect of CNV load, we derived three variables for each individual: the total number of CNVs that passed the QC criteria outlined above; the total length of these variants; and the number of genes disrupted by these CNVs. Genes were counted as ‘disrupted’ if there was any overlap between called CNV regions and known genetic co-ordinates +/−20 kb. The effect of CNV load on intelligence was investigated by fitting linear regression models to derived intelligence (g) factors and test scores. A number of regression models were fitted, using residualised gf and gc factor scores, corrected for age and sex effects, against total number of CNVs (rate), total CNV length, and the total number of genes disrupted, with ‘cohort’ fitted as a covariate.

CNV regions

Numerous copy-number variable regions have been implicated in mental health disorders. Williams et al. [54] identify 20 regions that have been associated with schizophrenia or autism spectrum disorders. We investigated the effect of these specific variants in our cohorts; i.e., we checked whether any individuals carried the disease-associated variants reported in autism spectrum disorders [45] or schizophrenia [51]. Where more than two individuals within the sample carried one of these variants, the differences in means of the carriers and non-carriers were tested using a t-test in R. Permutation analysis was performed to generate corrected p-values, on 100,000 permutations of the data. For each permutation, phenotypic values were permuted with reference to individual IDs, and a t-test was performed for each CNV region identified in Williams et al. [54]. The maximum test statistic of these 20 tests was retained at each permutation. Observed p-values were compared to the distribution of test statistics, with empirical p-values calculated for each region as the proportion of permutations where the maximum test statistic was greater than the observed statistic.


Following the QC procedures outlined above, 3133 individuals remained with phenotypic information for gf, and 3210 for gc. The CNVs used in subsequent analyses were those that fulfilled the QC procedures outlined above, and were called as CNVs by both QuantiSNP and PennCNV. For the samples providing the cognitive phenotypes gf, and gc, the total numbers of long, rare CNVs at ≥500 kb present in ≤1% of the cohort samples was 167 for both phenotypes. This gives overall CNV rates of 0.053 and 0.052 CNVs per individual for gf and gc (tables 1 and and2),2), with the slight discrepancy in total CNV counts due to different numbers of individuals with each phenotype. Most individuals carried no rare CNVs and, of those carrying any, the majority carried a single variant, with only nine individuals carrying more than one (Tables 3 and and4).4). Variants are observed across all autosomes throughout the sample, with more observed on longer chromosomes. The numbers of genes disrupted by these CNVs are shown in Tables 5 and and66 for gf and gc respectively, determined by comparing CNV boundaries to genetic start and end co-ordinates (+/−20 kb) and counting the number that overlap.

Table 1
Total CNV burden in each cohort for fluid-type intelligence (gf).
Table 2
Total CNV burden in each cohort for crystallized-type intelligence (gc).
Table 3
Distribution of long, rare CNVs in each cohort for individuals with a fluid-type intelligence (gf).
Table 4
Distribution of long, rare CNVs in each cohort for individuals with a crytallized-type intelligence (gc).
Table 5
Total genes disrupted by CNVs in each cohort for fluid-type intelligence (gf).
Table 6
Total genes disrupted by CNVs in each cohort for crystallized-type intelligence (gc).

A linear regression model was fitted on total CNV rate (number of CNVs called) per individual, fitting cohort as a covariate, summarised in Tables 7 and and88 for gf and gc respectively,. Results are also shown for similar regressions performed on the total length of CNVs per individual, and for the number of genes disrupted by CNVs. Models were fitted using data from all types of CNV, and for deletions and duplications separately. None of the fitted regression models were significant at p<0.05 for CNV effects. For numbers of rare CNVs, we grouped individuals into “carriers” and “non-carriers” and compared the g scores in these two groups. We found no significant differences between them, for all CNVs, or for deletions or duplications alone (data not shown).

Table 7
Tests of significance of CNV load on regression on fluid-type (gf) intelligence.
Table 8
Tests of significance of CNV load on regression on crystallized-type (gc) intelligence.

We also examined the effect of shorter, rare CNVs, at lengths 100–200 kb and 200–500 kb. These data are presented in Tables S1 and S2.Of these tests, the majority showed no association with either fluid or crystallized intelligence, but one test, total CNV counts for deletions in the range 200–500 kb, showed a nominally significant p-value of 0.039 with gf, although this is not robust to multiple testing correction. Yeo et al. [52] reported a significant association between rare deletions and Full-Scale IQ, but their definition of ‘rare’ (5%) differs from our 1% threshold, and other QC criteria differ. Repeating the analysis using the QC criteria of Yeo et al. on our samples failed to replicate their association (Tables S3 & S4).

Previous studies that have investigated the effect of rare CNVs on so-called neurocognitive disorders found several copy number variants that have an effect on these traits [39], [43], [51], [68][74]. Because these disorders also involve cognitive deficits, we examined variants declared significant in studies for autism and schizophrenia in our samples. The results are listed in Table S5. Where more than two individuals carried any particular variant, the sample was split into carriers and non-carriers, and differences in intelligence between these two groups were assessed using t-tests. Of the twenty loci examined, one region, 16p13.11, had four CNV carriers within our sample, and a region overlapping SHANK3 had three. The 16p13.11 region showed no evidence of an effect on either gf or gc, but SHANK3 showed a nominally significant effect at p = 0.006 for fluid-type intelligence. Permutation analyses were performed to generate corrected p-values: −100,000 permutations were performed, the largest test statistic taken over all 20 CNV regions, and the observed test statistic for each region compared to this distribution to calculate an empirical p-value. Following this procedure, SHANK3 remained significant for gf with a corrected p-value of 0.01.(Table S5).


Intelligence differences are substantially heritable, but studies to date have failed to find replicable associations between SNPs and cognitive traits that account for variation in intelligence in the normal population [8], [11]. One potential source of the missing heritability is copy-number variation. Following a similar approach to Williams et al. [54], in which the combined effect of rare CNVs on variation in ADHD was investigated in a sample of 366 cases and 1047 controls, we examined whether variation in rare CNVs had any effect within older cohorts with intelligence distributions in the normal range had any effect on the variation in cognitive ability. No significant combined effect of rare CNVs was found on intelligence in our combined sample of over 3,000 elderly individuals.

We found that the total load of CNVs longer than 500 kb per individual was not significantly associated with fluid- or crystallized-type intelligence phenotypes. Neither was total length of copy number variants, nor the total number of genes disrupted by rare CNVs. Testing for differences in intelligence between individuals carrying CNVs known to effect neurocognitive phenotypes, and non-carriers found a suggestive effect of SHANK3 on fluid intelligence scores. SHANK3 is a post-synaptic density protein involved in the regulation of synaptic transmission, and has been implicated in both autism and schizophrenia. SHANK3 is within the region of the chromosome 22q13.3 deletion syndrome, which is characterized by neonatal hypotonia, global developmental delay, severe cognitive deficits, normal to accelerated growth, absent to severely delayed speech, autistic behaviour, and minor dysmorphic features [73], [75]. Haploinsufficiency of this gene as a major causative factor in the neurologic symptoms of 22q13 deletion syndrome [76], and Gauthier et al. [77] identified two de novo mutations (R1117X and R536W) in two families with schizophrenia, in patients also displaying borderline or mild mental retardation. A recent GWAS study on cognitive phenotypes which used SNPs to tag common CNVs [78], found no significant association between these tagging markers and any of the measured cognitive phenotypes after correcting for multiple testing. This study by Need et al. [78] focussed on associations between cognitive phenotypes and specific copy-number variants associated with psychiatric illnesses in a sample of 1,000. Similarly, Saus et al. [79], using a candidate gene approach, found no significant differences between rare variants in cases and controls in major depressive disorder, bipolar disorder, schizophrenia or anxiety disorders, testing a control group of 341 individuals against case samples of ~200 per disorder.

Comparing our results to Yeo et al. [52], who found significant associations between rare deletions and variation in intelligence in a sample of 77 individuals, we failed to replicate their observed significant association between the extent of rare deletions and general intelligence, even when applying the same set of CNV QC criteria to our larger data set There are several possible reasons why this may be so. Whereas the phenotypic measures of intelligence that were used are broadly comparable, there are slight differences in our CNV calling and QC methods, the major one being our use of a minimum 500 kb cut-off in CNV length, but no such cut-off was used by Yeo et al. This is apparent from the overall rate of CNVs per individual in the two studies, with Yeo et al. finding an average of 17.42 CNVs per individual, with an average deletion rate of 10.95 (SD = 5.48) and insertion rate of 6.47 (SD = 9.82); compared to our observations of ~0.05 in total, a substantial disparity between the two rates, and between Yeo et al. and other studies of ‘rare’ CNVs which include a length cut-off [80]. The rationale behind such a cut-off is that the calls for longer CNVs are more accurate than shorter variants, and including short variants in an association analysis may increase the probability of type-I errors. When applying the QC criteria of Yeo et al. to our data (removing the length restriction, and excluding common CNVs at the 5% threshold), the observed average rate of CNVs called per individual was 15.48 (SD = 7.59). None of the regression analyses performed on CNVs called using these criteria showed a significant effect in our sample (Tables S3 and S4).

The samples studied in the present report comprise individuals within the normal cognitive range and we would expect the rate of CNVs detected to be similar to that obtained in other groups of healthy individuals that have undergone the same CNV quality control and selection procedures. Several studies have investigated the effect of long, rare CNVs on disease susceptibility, using the same length and frequency criteria employed here. Comparing our rates of CNV detection to the control groups of these studies, our observed values of 0.052 and 0.053 are comparable to the values of 0.05 observed by Blauw et al. [49], slightly lower than the value of 0.075 reported by Williams et al. [54], and lower than the values of 0.12 reported by the ISC [51], 0.17 in Pinto et al. [50], and 0.1924 in Bochukova et al. [81]. Some of these discrepancies can partially be accounted for by differences in genotype platforms and CNV calling algorithms. Blauw et al.[49] genotyped their control samples using a number of different platforms, but analysed only CNVs called by markers common to all: these were effectively the markers present on the HumanHap 300 array, comprising ~300 K SNP markers and lacking CNV specific probes. The ISC study [51] used genotypes derived from Affymetrix 5.0 and 6.0 arrays, and noted differences between these two arrays within their study. Bochukova et al. [81] also used the Affymetrix 6.0, whereas Pinto et al. [50] used the Illumina 1M chip for genotyping, and both observed a higher rate of CNVs per sample. Whenever a minimum number of SNPs is used as a criterion to define a CNV, there will be de facto more CNVs called on chips with higher marker densities.

Although the absolute rates of CNVs called differ between studies, the proportions of deletions detected compared to duplications are more consistent between studies. Of our total observed variants, 24.6% (41/167),are deletions in, compared to 28.0% in Bochukova et al. [81], 28.3% in ISC, and 29.2 % in Pinto et al [50]. The anomalously large proportion of 72.0% deletions in Blauw et al. [82] may be due to the set of markers used detecting more deletions. The 16.7% duplications called in Williams et al. [54] is smaller than other studies, perhaps due to the small sample size.

There are several other reasons why we may have failed to detect any effect of rare CNVs on cognitive ability in the cohorts analysed here. Primary amongst them, we are unlikely to have captured all of the genetic variation present within our samples. The Illumina 610 Quad chip used for genotyping in this study contains several non-polymorphic markers in known copy number variable regions, but will still not capture all of the variation present. With subsequent developments in microarrays, such as the Illumina 1M array used by Yeo et al., and the advent of reasonably priced whole-genome sequencing, we could capture a higher proportion of the actual variation. Other factors beyond genetic and structural variation may also contribute towards variation in general cognitive ability, including environmental variation and gene methylation or other epigenetic effects. Significant regions for rare CNVs reported in other neurocognitive disorders are not found significant by genome-wide associations using SNPs. Need et al. [78] suggest that looking for the effect of rare variants enriched in schizophrenia patients within a healthy population may reveal an association. Of the regions we examined, SHANK3 remained significant following permutation analysis, suggesting that this gene may be involved in normal variation in fluid intelligence. However, of the three CNV carriers identified in our samples that overlap the SHANK3 region, two carried duplications (copy number 3) and one a deletion (copy number 1), suggesting that imbalance rather than copy number per se may be important, but this counter-intuitive observation requires further investigation.

Many illnesses are associated with lower cognitive ability, and there is evidence that these states often involve structural genetic variation. Cognitive impairment is often a symptom in genomic syndromes associated with specific large-scale structural variation, including Williams-Beuren, Smith-Magenis, and Velo-Cardio-Facial Syndrome amongst others [83]. These syndromes are characterised by recurrent deletions or duplications at specific loci, which are large enough to be detected using Fluorescence In-Situ Hybridisation (FISH) or other microscopic techniques. Recent advances in microarray technology have allowed detection and characterisation of sub-microscopic structural variants, and found them to be ubiquitous throughout the genome. These copy-number variants are a major source of normal human genetic variation [84], but have also been found to be associated with complex disorders, including many neuro-psychiatric conditions, (for example, ADHD, depression, schizophrenia and autism-spectrum disorders).

To conclude, we find that, within the analytical limitations of the detection system available to us, there is no evidence for the effect of total CNV load on intelligence within the normal older population. Looking at specific CNV regions, we find evidence to suggest that copy number variation in the SHANK3,region where copy number variation has been previously associated with susceptibility to autism and schizophrenia, is associated with normal variation in fluid intelligence. However, our study does not preclude further contributions of CNVs at either extremes of the normal range of intelligence, or indeed on an individual by individual basis. New tools, including whole genome resequencing of individuals and their relatives with life-course measures of intelligence would be valuable in further resolving the important issue of identifying genetic contributions to individual differences in intelligence.

Supporting Information

Table S1

Tests of significance of CNV load on regression on fluid-type intelligence (gf).


Table S2

Tests of significance of CNV load on regression on crystallized-type intelligence (gc).


Table S3

Tests of significance of CNV load on regression on fluid-type (gf) intelligence for rare CNVs present at ≤ 5% frequency in each cohort, with no length restriction.


Table S4

Tests of significance of CNV load on regression on fluid-type (gf) intelligence for rare CNVs present at ≤ 5% frequency in each cohort, with no length restriction.


Table S5

Significance of individual CNV loci previously implicated in psychiatric disorders for fluid-type (gf) and crystallized-type (gc) intelligence.



We thank the cohort participants and research teams who contributed to these studies.

Funding Statement

Genotyping and analyses of the cohorts conducted here were supported by the UK's Biotechnology and Biological Sciences Research Council (BBSRC). ML is a Royal Society of Edinburgh/Lloyds TSB Foundation for Scotland Personal Research Fellow. PAT is partially funded by an RCUK Fellowship. LML is funded by the AXA research fund. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.


1. Deary IJ (2008) Why do intelligent people live longer? Nature 456: 175–176 [PubMed]
2. Deary IJ, Penke L, Johnson W (2010) The neuroscience of human intelligence differences. Nature Reviews Neuroscience 11: 201–211 [PubMed]
3. Spearman C (1904) “General intelligence” objectively determined and measured. American Journal of Psychology 15: 201–292 [PubMed]
4. Carroll JB (1993) Human cognitive abilities: A survey of factor-analytical studies. New York: Cambridge University Press.
5. Johnson W, te Nijenhuis J, Bouchard TJ (2008) Still just 1 g: Consistent results from five test batteries. Intelligence 36: 81–95
6. Deary IJ, Whiteman MC, Starr JM, Whalley LJ, Fox HC (2004) The impact of childhood intelligence on later life: Following up the Scottish Mental Surveys of 1932 and 1947. Journal of Personality and Social Psychology 86: 130–147 [PubMed]
7. Salthouse TA (2004) Localizing age-related individual differences in a hierarchical structure. Intelligence 32: 541–561 [PMC free article] [PubMed]
8. Deary IJ, Johnson W, Houlihan LM (2009) Genetic foundations of human intelligence. Human Genetics 126: 215–232 [PubMed]
9. Davies G, Tenesa A, Payton A, Yang J, Harris SE, et al. (2011) Genome-wide association studies establish that human intelligence is highly heritable and polygenic. Molecular Psychiatry 16: 996–1005 [PMC free article] [PubMed]
10. Houlihan LM, Harris SE, Luciano M, Gow AJ, Starr JM, et al. (2009) Replication study of candidate genes for cognitive abilities: the Lothian Birth Cohort 1936. Genes Brain and Behavior 8: 238–247 [PubMed]
11. Payton A (2009) The Impact of Genetic Research on our Understanding of Normal Cognitive Ageing: 1995 to 2009. Neuropsychology Review 19: 451–477 [PubMed]
12. Visscher PM (2008) Sizing up human height variation. Nature Genetics 40: 489–490 [PubMed]
13. Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, et al. (2009) Finding the missing heritability of complex diseases. Nature 461: 747–753 [PMC free article] [PubMed]
14. Eichler EE, Flint J, Gibson G, Kong A, Leal SM, et al. (2010) VIEWPOINT Missing heritability and strategies for finding the underlying causes of complex disease. Nature Reviews Genetics 11: 446–450 [PMC free article] [PubMed]
15. Yang JA, Benyamin B, Mcevoy BP, Gordon S, Henders AK, et al. (2010) Common SNPs explain a large proportion of the heritability for human height. Nature Genetics 42: 565–U131 [PMC free article] [PubMed]
16. Mitchell KJ, Porteous DJ (2010) Rethinking the Genetic Architecture of Schizophrenia. Schizophrenia Research 117: 222
17. Cordell HJ (2009) Detecting gene-gene interactions that underlie human diseases. Nature Reviews Genetics 10: 392–404 [PMC free article] [PubMed]
18. Slatkin M (2009) Epigenetic Inheritance and the Missing Heritability Problem. Genetics 182: 845–850 [PubMed]
19. Feuk L, Carson AR, Scherer SW (2006) Structural variation in the human genome. Nature Reviews Genetics 7: 85–97 [PubMed]
20. Redon R, Ishikawa S, Fitch KR, Feuk L, Perry GH, et al. (2006) Global variation in copy number in the human genome. Nature 444: 444–454 [PMC free article] [PubMed]
21. Schrider DR, Hahn MW (2010) Gene copy-number polymorphism in nature. Proceedings of the Royal Society B-Biological Sciences 277: 3213–3221 [PMC free article] [PubMed]
22. Iafrate AJ, Feuk L, Rivera MN, Listewnik ML, Donahoe PK, et al. (2004) Detection of large-scale variation in the human genome. Nature Genetics 36: 949–951 [PubMed]
23. Beckmann JS, Estivill X, Antonarakis SE (2007) Copy number variants and genetic traits: closer to the resolution of phenotypic to genotypic variability. Nature Reviews Genetics 8: 639–646 [PubMed]
24. Carter NP (2007) Methods and strategies for analyzing copy number variation using DNA microarrays. Nature Genetics 39: S16–S21 [PMC free article] [PubMed]
25. Dellinger AE, Saw SM, Goh LK, Seielstad M, Young TL et al. . (2010) Comparative analyses of seven algorithms for copy number variant identification from single nucleotide polymorphism arrays. Nucleic Acids Research 38: [PMC free article] [PubMed]
26. Zhang D, Qian Y, Akula N, Alliey-Rodriguez N, Tang J et al. . (2011) Accuracy of CNV Detection from GWAS Data. PLOS ONE. [PMC free article] [PubMed]
27. Gonzalez E, Kulkarni H, Bolivar H, Mangano A, Sanchez R, et al. (2005) The influence of CCL3L1 gene-containing segmental duplications on HIV-1/AIDS susceptibility. Science 307: 1434–1440 [PubMed]
28. McKinney C, Merriman ME, Chapman PT, Gow PJ, Harrison AA, et al. (2008) Evidence for an influence of chemokine ligand 3-like 1 (CCL3L1) gene copy number on susceptibility to rheumatoid arthritis. Annals of the Rheumatic Diseases 67: 409–413 [PubMed]
29. Yang Y, Chung EK, Wu YL, Nagaraja HN, Zhou B, et al. (2007) Complement C4 gene copy number variation in human autoimmune disease systemic lupus erythematosus (SLE). Molecular Immunology 44: 261
30. Fellermann K, Stange DE, Schaeffeler E, Schmalzl H, Wehkamp J, et al. (2006) A chromosome 8 gene-cluster polymorphism with low human beta-defensin 2 gene copy number predisposes to Crohn disease of the colon. American Journal of Human Genetics 79: 439–448 [PubMed]
31. Hollox EJ, Huffmeier U, Zeeuwen PLJM, Palla R, Lascorz J, et al. (2008) Psoriasis is associated with increased beta-defensin genomic copy number. Nature Genetics 40: 23–25 [PMC free article] [PubMed]
32. Frank B, Hemminki K, Meindl A, Wappenschmidt B, Sutter C et al. . (2007) BRIP1 (BACH1) variants and familial breast cancer risk: a case-control study. Bmc Cancer 7: [PMC free article] [PubMed]
33. Merikangas AK, Corvin AP, Gallagher L (2009) Copy-number variants in neurodevelopmental disorders: promises and challenges. Trends in Genetics 25: 536–544 [PubMed]
34. Rovelet-Lecruz A, Hannequin D, Raux G, Le Meur N, Laquerriere A, et al. (2006) APP locus duplication causes autosomal dominant early-onset Alzheimer disease with cerebral amyloid angiopathy. Nature Genetics 38: 24–26 [PubMed]
35. Ibanez P, Bonnet AM, Debarges B, Lohmann E, Tison F, et al. (2004) Causal relation between alpha-synuclein gene duplication and familial Parkinson's disease. Lancet 364: 1169–1171 [PubMed]
36. Singleton AB, Farrer M, Johnson J, Singleton A, Hague S, et al. (2003) alpha-synuclein locus triplication causes Parkinson's disease. Science 302: 841. [PubMed]
37. Helbig I, Mefford HC, Sharp AJ, Guipponi M, Fichera M, et al. (2009) 15q13.3 microdeletions increase risk of idiopathic generalized epilepsy. Nature Genetics 41: 160–162 [PMC free article] [PubMed]
38. Xu B, Roos JL, Levy S, Van Rensburg EJ, Gogos JA, et al. (2008) Strong association of de novo copy number mutations with sporadic schizophrenia. Nature Genetics 40: 880–885 [PubMed]
39. Stefansson H, Rujescu D, Cichon S, Pietilainen OPH, Ingason A, et al. (2008) Large recurrent microdeletions associated with schizophrenia. Nature 455: 232–U61 [PMC free article] [PubMed]
40. McMullan DJ, Bonin M, Hehir-Kwa JY, de Vries BBA, Dufke A, et al. (2009) Molecular Karyotyping of Patients with Unexplained Mental Retardation by SNP Arrays: A Multicenter Study. Human Mutation 30: 1082–1092 [PubMed]
41. Edelmann L, Hirschhorn K (2009) Clinical Utility of Array CGH for the Detection of Chromosomal Imbalances Associated with Mental Retardation and Multiple Congenital Anomalies. Year in Human and Medical Genetics 2009 1151: 157–166 [PubMed]
42. Bijlsma EK, Gijsbers ACJ, Schuurs-Hoeijmakers JHM, van Haeringen A, van de Putte DEF, et al. (2009) Extending the phenotype of recurrent rearrangements of 16p11.2: Deletions in mentally retarded patients without autism and in normal individuals. European Journal of Medical Genetics 52: 77–87 [PubMed]
43. Weiss LA, Shen YP, Korn JM, Arking DE, Miller DT, et al. (2008) Association between microdeletion and microduplication at 16p11.2 and autism. New England Journal of Medicine 358: 667–675 [PubMed]
44. Szatmari P, Paterson AD, Zwaigenbaum L, Roberts W, Brian J, et al. (2007) Mapping autism risk loci using genetic linkage and chromosomal rearrangements. Nature Genetics 39: 319–328 [PubMed]
45. Marshall CR, Noor A, Vincent JB, Lionel AC, Feuk L, et al. (2008) Structural variation of chromosomes in autism spectrum disorder. American Journal of Human Genetics 82: 477–488 [PubMed]
46. Kumar RA, KaraMohamed S, Sudi J, Conrad DF, Brune C, et al. (2008) Recurrent 16p11.2 microdeletions in autism. Human Molecular Genetics 17: 628–638 [PubMed]
47. Glessner JT, Wang K, Sleiman PMA, Zhang H, Kim CE et al. . (2010) Duplication of the SLIT3 Locus on 5q35.1 Predisposes to Major Depressive Disorder. PLOS ONE 5: [PMC free article] [PubMed]
48. Itsara A, Cooper GM, Baker C, Girirajan S, Li J, et al. (2009) Population Analysis of Large Copy Number Variants and Hotspots of Human Genetic Disease. American Journal of Human Genetics 84: 148–161 [PubMed]
49. Blauw HM, Al Chalabi A, Andersen PM, van Vught PWJ, Diekstra FP, et al. (2010) A large genome scan for rare CNVs in amyotrophic lateral sclerosis. Human Molecular Genetics 19: 4091–4099 [PubMed]
50. Pinto D, Pagnamenta AT, Klei L, Anney R, Merico D, et al. (2010) Functional impact of global rare copy number variation in autism spectrum disorders. Nature 466: 368–372 [PMC free article] [PubMed]
51. Stone JL, O'Donovan MC, Gurling H, Kirov GK, Blackwood DHR, et al. (2008) Rare chromosomal deletions and duplications increase risk of schizophrenia. Nature 455: 237–241 [PMC free article] [PubMed]
52. Yeo RA, Gangestad SW, Liu J, Calhoun VD, Hutchison KE (2011) Rare Copy Number Deletions Predict Individual Variation in Intelligence. PLOS ONE 6: e16339. [PMC free article] [PubMed]
53. Wechsler D (1999) Wechsler Abbreviated Scale of Intelligence. San Antonio: Psychological Corporation.
54. Williams NM, Zaharieva I, Martin A, Mantripragada K, Fossdal R, et al. (2010) Rare chromosomal deletions and duplications in attention-deficit hyperactivity disorder: a genome-wide analysis. The Lancet 376: 1401–1408 [PMC free article] [PubMed]
55. Hill SK, Keshavan MS, Thase ME, Sweeney JA (2004) Neuropsychological dysfunction in anti psychotic-naive first-episode unipolar psychotic depression. American Journal of Psychiatry 161: 996–1003 [PubMed]
56. Deary IJ, Whalley LJ, Starr JM (2009) A Lifetime of Intelligence: Follow-up Studies of the Scottish Mental Surveys of 1932 and 1947. Washington D.C: American Psychological Association.
57. Scottish Council for Research in Education (1933) The Intelligence of Scottish Children: A National Survey of an Age-group. London: University of London Press.
58. Raven JC, Court JH, Raven J. (1977) Manual for Raven's Progressive Matrices and Vocabulary Scales. London: H.K.Lewis.
59. Lezak MD, Howieson D, Loring D.W. (2004) Neuropsychological Assessment. Oxford: Oxford University Press.
60. Wechsler D (1998) WAIS-III UK Administration and Scoring Manual. New York: Psychological Corporation.
61. Nelson HE, Willison JR (1991) National Adult Reading Test (NART) Test Manual. Windsor: NFER-Nelson.
62. Deary IJ, Gow AJ, Taylor MD, Corley J, Brett C et al. . (2007) The Lothian Birth Cohort 1936: a study to examine influences on cognitive ageing from age 11 to age 70 and beyond. BMC Geriatrics 7: [PMC free article] [PubMed]
63. Wechsler D (1981) Wechsler Adult Intelligence Scale-Revised. London: Psychological Corporation.
64. Rabbitt PMA, McInnes L, Diggle P, Holland F, Bent N, et al. (2004) The University of Manchester longitudinal study of cognition in normal healthy old age, 1983 through 2003. Aging Neuropsychology and Cognition 11: 245–279
65. Johnson W, Bouchard TJ, Krueger RF, Mcgue M, Gottesman II (2004) Just one g: consistent results from three test batteries. Intelligence 32: 95–107
66. Wang K, Li MY, Hadley D, Liu R, Glessner J, et al. (2007) PennCNV: An integrated hidden Markov model designed for high-resolution copy number variation detection in whole-genome SNP genotyping data. Genome Research 17: 1665–1674 [PubMed]
67. Colella S, Yau C, Taylor JM, Mirza G, Butler H, et al. (2007) QuantiSNP: an Objective Bayes Hidden-Markov Model to detect and accurately map copy number variation using SNP genotyping data. Nucleic Acids Research 35: 2013–2025 [PMC free article] [PubMed]
68. Glessner JT, Wang K, Cai GQ, Korvatska O, Kim CE, et al. (2009) Autism genome-wide copy number variation reveals ubiquitin and neuronal genes. Nature 459: 569–573 [PMC free article] [PubMed]
69. Ingason A, Rujescu D, Cichon S, Sigurdsson E, Sigmundsson T et al. . (2009) Copy number variations of chromosome 16p13.1 region associated with schizophrenia. Mol Psychiatry. [PMC free article] [PubMed]
70. Kalscheuer VM, FitzPatrick D, Tommerup N, Bugge M, Niebuhr E, et al. (2007) Mutations in autism susceptibility candidate 2 (AUTS2) in patients with mental retardation. Human Genetics 121: 501–509 [PubMed]
71. Kim HG, Kishikawa S, Higgins AW, Seong IS, Donovan DJ, et al. (2008) Disruption of neurexin 1 associated with autism spectrum disorder. American Journal of Human Genetics 82: 199–207 [PubMed]
72. McCarthy SE, Makarov V, Kirov G, Addington AM, McClellan J, et al. (2009) Microduplications of 16p11.2 are associated with schizophrenia. Nature Genetics 41: 1223–1U85 [PMC free article] [PubMed]
73. Moessner R, Marshall CR, Sutcliffe JS, Skaug J, Pinto D, et al. (2007) Contribution of SHANK3 mutations to autism spectrum disorder. American Journal of Human Genetics 81: 1289–1297 [PubMed]
74. Roohi J, Montagna C, Tegay DH, Palmer LE, DeVincent C, et al. (2009) Disruption of contactin 4 in three subjects with autism spectrum disorder. Journal of Medical Genetics 46: 176–182 [PMC free article] [PubMed]
75. Durand CM, Betancur C, Boeckers TM, Bockmann J, Chaste P, et al. (2007) Mutations in the gene encoding the synaptic scaffolding protein SHANK3 are associated with autism spectrum disorders. Nature Genetics 39: 25–27 [PMC free article] [PubMed]
76. Wilson HL, Wong ACC, Shaw SR, Tse WY, Stapleton GA, et al. (2003) Molecular characterisation of the 22q13 deletion syndrome supports the role of haploinsufficiency of SHANK3/PROSAP2 in the major neurological symptoms. Journal of Medical Genetics 40: 575–584 [PMC free article] [PubMed]
77. Gauthier J, Siddiqui T, Huashan P, Yokomaku D, Hamdan F, et al. (2011) Truncating mutations in NRXN2 and NRXN1 in autism spectrum disorders and schizophrenia. Human Genetics 130: 563–573 [PMC free article] [PubMed]
78. Need AC, Attix DK, Mcevoy JM, Cirulli ET, Linney KL, et al. (2009) A genome-wide study of common SNPs and CNVs in cognitive performance in the CANTAB. Human Molecular Genetics 18: 4650–4661 [PMC free article] [PubMed]
79. Saus E, Brunet A, Armengol L, Alonso P, Crespo JM, et al. (2010) Comprehensive copy number variant (CNV) analysis of neuronal pathways genes in psychiatric disorders identifies rare variants within patients. Journal of Psychiatric Research 44: 971–978 [PubMed]
80. Wang K, Li MY, Hadley D, Liu R, Glessner J, et al. (2007) PennCNV: An integrated hidden Markov model designed for high-resolution copy number variation detection in whole-genome SNP genotyping data. Genome Research 17: 1665–1674 [PubMed]
81. Bochukova EG, Huang N, Keogh J, Henning E, Purmann C, et al. (2010) Large, rare chromosomal deletions associated with severe early-onset obesity. Nature 463: 666–670 [PMC free article] [PubMed]
82. Blauw HM, Veldink JH, van Es MA, van Vught PW, Saris CG, et al. (2008) Copy-number variation in sporadic amyotrophic lateral scleroses: a genome-wide screen. Lancet Neurology 7: 319–326 [PubMed]
83. Morrow EM (2010) Genomic copy number variation in disorders of cognitive development. J Am Acad Child Adolesc Psychiatry 49: 1091–1104 [PMC free article] [PubMed]
84. McCarroll SA, Kuruvilla FG, Korn JM, Cawley S, Nemesh J, et al. (2008) Integrated detection and population-genetic analysis of SNPs and copy number variation. Nature Genetics 40: 1166–1174 [PubMed]

Articles from PLoS ONE are provided here courtesy of Public Library of Science