Protocol to identify functional doppelgängers and verify biomedical gene expression data using...

Area of Science:

• Computational Biology
• Machine Learning
• Bioinformatics

Background:

• Independently derived sample pairs, known as functional doppelgängers (FDs), can inadvertently be split between training and validation datasets.
• This misassignment of FDs can lead to inflated or underestimated machine learning (ML) model performance metrics.
• Accurate identification of FDs is crucial for reliable ML model development in biomedical research.

Purpose of the Study:

• To introduce doppelgangerIdentifier (DI), a novel software tool for the identification of functional doppelgängers.
• To provide a comprehensive protocol for the installation, data preparation, and application of DI.
• To demonstrate the utility of DI using real-world biomedical gene expression data.

Main Methods:

• Software development of doppelgangerIdentifier (DI) for automated FD detection.
• Detailed guidelines for data preprocessing and input formatting compatible with DI.
• Functional testing and validation of DI's performance on benchmark datasets, including gene expression data.

Main Results:

• Successful implementation of DI software, enabling efficient identification of FDs.
• Demonstration of DI's capability to accurately detect FDs in complex biomedical datasets.
• Provision of user-friendly instructions and parameter selection guidance for DI.

Conclusions:

• The doppelgangerIdentifier (DI) software provides a robust solution for detecting functional doppelgängers in biological data.
• Implementing DI can significantly enhance the reliability and reproducibility of machine learning models in biomedical applications.
• This work offers a critical tool for researchers aiming to mitigate confounding factors in ML-driven biological discovery.