Simple method for cutoff point identification in descriptive high-throughput biological studies.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics
• Proteomics

Background:

• High-throughput omics technologies generate vast datasets, necessitating robust methods for data analysis.
• Existing cutoff identification algorithms are unsuitable for descriptive studies lacking predefined biological associations.
• There is a need for novel approaches to analyze continuous variables in omics research.

Purpose of the Study:

• To develop and present an accessible algorithm for cutoff point identification in descriptive omics studies.
• To address the limitations of current methods in scenarios without known biological variable associations.
• To facilitate the prioritization of significant variables for downstream functional analysis.

Main Methods:

• The study proposes an algorithm based on the biphasic distribution commonly observed in ranked continuous omics variables.
• The algorithm identifies the boundary between a slow-growing initial phase and a rapid-growing second phase.
• This boundary serves as the cutoff point for variable selection.

Main Results:

• The ranked distribution of continuous variables in high-throughput descriptive studies often exhibits a biphasic curve.
• The developed algorithm successfully identifies the transition point in this biphasic distribution.
• Application to diverse datasets (human genes, mammalian genes, human proteins) yielded relevant variable shortlists.

Conclusions:

• The algorithm assumes that a small subset of high-value variables significantly influences biological system functions.
• Tested on human and mammalian omics data, the method effectively identified key genes and proteins relevant to dominant biological functions.
• This cutoff identification method offers a valuable tool for prioritizing variables in descriptive omics studies when other methods are not applicable.