Mitigating false-positive associations in rare disease gene discovery

Area of Science:

• Genomics
• Rare Diseases
• Genetic Association Studies

Background:

• Clinical sequencing is advancing, yet causal variants remain unidentified in most rare disease cases.
• Unsolved cases are valuable for gene discovery, particularly when similar phenotypes are found across individuals.
• Existing systems like Matchmaker Exchange facilitate identifying individuals with shared phenotypes.

Purpose of the Study:

• To describe and quantify the risks of false-positive associations in gene discovery from unsolved rare disease cases.
• To develop a method for calculating the statistical significance of gene-phenotype matches.
• To estimate the number of cases required for gene discovery in various disorders.

Main Methods:

• Utilized variant data from the Exome Aggregation Consortium (over 60,000 individuals) to establish prior probabilities of gene variants.
• Developed a statistical model to calculate P values for gene-phenotype matches based on cohort size, variant frequency, gene, and inheritance mode.
• Applied the model to simulated rare disease cohorts, including examples like MECP2 and TTN.

Main Results:

• Demonstrated that a match in two of 10 patients for MECP2 is statistically significant (P = 0.0014).
• Showed that a similar match in TTN would not reach statistical significance (P > 0.999) due to higher variant frequency.
• Analyzed the probability of matches in clinical exome data to estimate case numbers needed for gene discovery.

Conclusions:

• The study quantifies the risk of false-positive associations in rare disease gene discovery efforts.
• A statistical framework and an online tool (Rare Disease Match) are provided to mitigate uncertainties in identifying causal genes.
• This approach aids in prioritizing candidate genes and optimizing gene discovery in unsolved rare disease cases.