Enhancing Classification of liquid chromatography mass spectrometry data with Batch Effect Removal Neural Networks...

Area of Science:

• Proteomics
• Computational Biology
• Analytical Chemistry

Background:

• Liquid Chromatography Mass Spectrometry (LC-MS) is vital for biological sample profiling.
• Batch effects from experimental variations hinder data interpretability and reproducibility.
• Existing batch effect correction methods often compromise genuine biological signals.

Approach:

• We introduce Batch Effect Removal Neural Networks (BERNN), a novel deep learning approach.
• BERNN models are designed to maximize sample classification performance across conditions.
• Emphasis is placed on efficient generalization to unseen experimental batches.

Key Points:

• BERNN models demonstrated superior sample classification performance across diverse LC-MS datasets.
• The best classification model did not always yield the most effective batch effect removal.
• Overcorrection of batch effects can lead to the loss of essential biological variability.

Conclusions:

• BERNN offers a promising solution for mitigating batch effects in large-scale LC-MS data.
• Balancing effective batch effect removal with the preservation of biological diversity is critical.
• This work advances reproducible proteomics research through improved data processing techniques.