Quantify and control reproducibility in high-throughput experiments.

Area of Science:

• Biomedical Research
• Computational Biology
• Genomics

Background:

• Reproducibility is essential for high-throughput experiments in biomedical research.
• Existing methods may not fully capture the nuances of reproducibility in complex datasets.

Purpose of the Study:

• To propose INTRIGUE, a novel set of computational methods for evaluating and controlling reproducibility in high-throughput experiments.
• To introduce a new definition of reproducibility emphasizing directional consistency using signed effect sizes.

Main Methods:

• Development of computational methods (INTRIGUE) to assess overall study reproducibility and individual experimental unit reproducibility.
• Utilizing signed effect size estimates to define and measure directional consistency.
• Simulation studies to detect unobserved batch effects.

Main Results:

• Demonstrated the ability of INTRIGUE to detect unobserved batch effects in simulations.
• Illustrated the application of INTRIGUE in transcriptome-wide association studies (TWAS).
• Showcased INTRIGUE's utility for quality control and investigating biological heterogeneity.

Conclusions:

• INTRIGUE provides a robust framework for assessing and enhancing reproducibility in high-throughput experiments.
• The methods are versatile, applicable to quality control and biological discovery in genomics.
• Potential extensions exist for addressing publication bias and conceptual replications in research.