Although individual genetic markers from genome-wide association studies (GWAS) usually explain limited heritability especially in psychiatric traits, polygenic scores can be a better way to summarise genetic effects of a large number of markers that do not individually achieve geneome-wide significance. (Dudbridge, 2013). This approach has been applied succesfully to schizophrenia (Purcell et al, 2009) and bipolar disorder (Hamshere et al, 2011).

Attention-deficit hyperactivity disorder (ADHD) is a highly heritable neurodevelopmental disorder. Traits of ADHD are present in family members of those with the conditions, as well as in the general population. Learning disabilities are common in ADHD with up to 40% of children with a diagnosis of ADHD presenting with them (Willcutt et al, 2000). In addition, neurocognitive deficits are seen in ADHD with executive function being the primary deficit (Willcutt et al, 2000). Given the attention problems that children with ADHD face, it is expected that their school performance will be affected.

Our hypothesis is that polygenic risk scores based on a clinical sample of ADHD will predict educational attainment and IQ in children from the general polulation.

Polygenic risk scores were calcualted on ALSPAC children using a GWAS of 727 ADHD patients and 5,081 controls (Stergiakouli et al, 2012) as a discovery sample (project B1342) and were provided by colleagues in Cardiff.

Aims of proposal:

To test if polygenic risk scores predict school performance in ALSPAC children using SATS school scores. We are expecting that higher polygenic scores for ADHD will be associated with worse performance in school.

To test if polygenic risk scores predict IQ in ALSPAC children using the WASI IQ score from Focus@3 clinic. We are expecting that higher polygenic scores for ADHD will be associated with lower IQ scores at age 15.

The training sample GWAS from Cardiff did not identify any genome-wide significant hits but it did find that 13 biological pathways enriched for SNP association significantly overlapped with those enriched for rare CNVs. These included cholesterol-related pathways. At the level of individual genes, CHRNA7, which encodes a nicotinic receptor subunit previously implicated in neuropsychiatric disorders, was affected by six large duplications in case subjects (none in comparison subjects) (Stergiakouli et al, 2012). Our hypothesis is that polygenic risk scores based on genetic variants from the GWAS mentioned above could predict obesity and smoking behaviour in the general population.

Aims of proposal:

To test if polygenic risk scores predict BMI at ages 7, 9 and 15 and lipid levels at age 9.

To test if polygenic risk scores predict smoking behaviour at age 15.