RSim: A reference-based normalization method via rank similarity

Area of Science:

• Microbiology
• Bioinformatics
• Computational Biology

Background:

• Microbiome sequencing data normalization is essential for accurate analysis.
• High frequencies of zero counts present a significant challenge in microbiome data normalization.
• Existing methods may introduce bias when handling zero counts.

Purpose of the Study:

• To introduce a novel reference-based normalization method, Normalization via Rank Similarity (RSim).
• To address the challenge of zero counts in microbiome data normalization.
• To improve the accuracy and robustness of downstream microbiome analyses.

Main Methods:

• Proposed a novel reference-based normalization method called Normalization via Rank Similarity (RSim).
• RSim corrects sample-specific biases without requiring additional assumptions or treatments for zero counts.
• Evaluated RSim's performance using numerical experiments.

Main Results:

• RSim effectively corrects sample-specific biases, even with a high prevalence of zero counts.
• The method reduces false discoveries and enhances detection power in downstream analyses.
• RSim improves the clarity of biological signals in Principal Coordinate Analysis (PCoA) plots, association analyses, and differential abundance analyses.

Conclusions:

• RSim offers a robust and unbiased approach to normalizing microbiome sequencing data.
• The method's ability to handle zero counts makes it suitable for diverse microbiome datasets.
• RSim facilitates more reliable biological interpretations from microbiome sequencing studies.