Genetic association study in psychiatry: analytical evaluation and a recommendation

Area of Science:

• Genetics
• Psychiatric Research
• Statistical Genetics

Background:

• Candidate gene association studies for psychiatric disorders often yield false positives due to numerous loci and low prior probability of relevance.
• Previous recommendations suggested lowering the statistical significance threshold (alpha level) to mitigate this issue.

Purpose of the Study:

• To evaluate the effectiveness of a fixed, predetermined alpha level adjustment for reducing false positives in genetic association studies.
• To demonstrate that false positive rates are influenced by various parameters beyond the alpha level.

Main Methods:

• The study analyzed the impact of disorder prevalence, genotype prevalence, and relative risk on false positive and false negative rates.
• It assessed scenarios where even multiple independent replications failed to adequately control false positive rates.

Main Results:

• Lowering the alpha level to a fixed value is not a recommended strategy for controlling false positives.
• False positive probabilities are highly dependent on specific parameters, with some situations showing rates exceeding 80-90% even with replications.
• The required P levels to minimize false positives can vary significantly between statistical tests.

Conclusions:

• A fixed alpha level adjustment is insufficient for managing false positives in genetic association studies.
• A flexible procedure for adjusting probability levels based on specific study parameters is necessary.
• This approach aims to improve the reliability of findings in psychiatric genetics research.