Candidate gene scoring to predict broad adolescent psychopathology

Candidate gene effects consistently fail to replicate. However, because it is now known that most genetic effects are incredibly minute, samples of the size typically employed in psychological research were undoubtedly too small to detect the effects of individual candidate genes. In addition, research showing strong genetic correlation among mental disorders suggests data on multiple disorders and their symptoms is the most appropriate for uncovering the etiology of mental illness. That is, single gene, single disorder studies are underpowered. We tested whether the combined effect of 121 candidate genes was sufficient to predict psychopathology in a sample of 343 adolescents. A genetic risk score was created with highly precise effect estimates from a genome-wide association study (GWAS) on 337,199 people. To maximize the strength of this score, we used transdiagnostic p-factor model scores as our measure of psychopathology. The genetic risk scores failed to predict in our sample and were dwarfed by age and gender effects, mirroring the genes’ weak and mostly non-significant results in the GWAS. Our results are most consistent with the view that the candidate gene approach is obsolete. However, modern molecular genetics studies like GWAS currently lack detailed, thorough phenotype measurement. Future work should focus on developing high quality, deeply phenotyped data currently lacking in large consortia efforts