Optimized phenotype definitions boost GWAS power

Area of Science:

• Genetics and genomics
• Computational biology
• Epidemiology

Background:

• Complex diseases pose significant health challenges, with genetics contributing substantially to disease risk.
• Observational data offers large-scale, cost-effective insights but is confounded by healthcare and societal factors.
• Purifying the genetic signal within observational data is crucial for accurate genetic discovery.

Purpose of the Study:

• Introduce MaxGCP, a novel phenotyping method to isolate the genetic component in observational data.
• Optimize phenotype definitions to maximize coheritability with complex traits of interest.
• Enhance the power of genetic studies using large-scale observational datasets.

Main Methods:

• MaxGCP defines phenotypes to maximize coheritability with the target complex trait.
• The method exhibits linear computational complexity, enabling scalability with numerous features.
• MaxGCP does not require manual feature selection, streamlining the phenotyping process.

Main Results:

• MaxGCP significantly enhances the genetic signal in observational data.
• In stroke analysis, MaxGCP increased study power by over 13% compared to traditional methods.
• The method demonstrates improved performance over conventional single-code phenotype definitions.

Conclusions:

• MaxGCP is an effective tool for improving genetic discovery in observational data.
• The method offers a phenotype-specific approach to isolate genetic signals.
• MaxGCP is anticipated to be widely applicable for studying complex diseases in observational datasets.