Our group has already completed a three-stage genomewide association study of mathematical ability in 10 year-olds drawn from the Twins Early Development Study (TEDS) (under review). 43 SNPs were nominated from the first two stages which involved assaying pooled DNA on Affymetrix DNA microarrays to compare individuals of high mathematical ability to those of low mathematical ability. In the third stage, these 43 SNPs were individually genotyped in a sample which spanned the entire spectrum of ability. 10 SNP associations remained significant. Though effect sizes were extremely small, when a composite SNP-score was created for subjects in the final sample by summing the number of 'mathemtics score increasing alleles', these 10 SNPs accounted for 2.9% of the phenotypic variation in the sample. Interestingly, when the remaining 33 SNPs were added to this score, they did not mask this effect on mathematical performance, but rather increased it. This suggests to us that many of these remaining 33 SNPs may show true influences over mathematical ability which when analysed alone, our sample was underpowered to detect.

As the true test of association is replication, we are currently seeking out a comparable sample with mathematics ability data. With access to 1) existing ALSPAC mathematics scores, and 2) existing genotype data - collected on Illumina microarrays - we would be able to simply and easily test for the association of the 43 implicated SNPs (or where necessary, their proxies on the Illumina platform) with mathematical ability in the ALSPAC sample. As quantitative genetic results indicate that some of the genetic action over mathematics is general across ages, whilst some is not, we propose to investigate the action of the 43 SNPs over mathematical ability at all ages for which there is data available (we have been informed that this is 4-5, 11 and 15 years). Further to this, we are especially interested in investigating the combined influence of 1) the top 10 SNPs and 2) all 43 SNPs over mathematical ability. As mentioned above, this may be achieved by simply summing the math-increasing alleles possesed by each subject, and investigating the relationship of this SNP-set score with mathematical ability. If linear models are used to investigate SNP associations, then a simple alternative might be to add all 10/43 SNPs into the model. Whilst we would greatly prefer to conduct these analyses ourselves - we do understand that there may be restrictions on the release of some or all of the data we have requested, and so would also be very happy to work closely with ALSPAC researchers on this project.

As much of the work our group conducts is funded by the Wellcome Trust, we follow the same open access publishing policy as ALSPAC. We will therefore ensure that any publications resulting from this collaboration will be freely available to all researchers via PubMed Central.