categoryCompare, an analytical tool based on feature annotations.

Area of Science:

• Computational biology and bioinformatics
• Systems biology
• Genomics, transcriptomics, proteomics, and metabolomics

Background:

• High-throughput omics data analysis typically focuses on individual features (genes, proteins, metabolites) or requires identical features across datasets for comparison.
• Existing comparative and meta-analysis methods struggle with data heterogeneity from different platforms and biological variability, hindering robust biomarker discovery.
• Common biological processes or pathways are often implicated across studies, despite variations in measured features.

Purpose of the Study:

• To develop and assess a novel methodology, categoryCompare, for cross-platform and cross-sample comparison of high-throughput omics data at the biological annotation level.
• To evaluate the utility of annotation-level comparison against feature-level comparison using hypothetical data.
• To investigate similarities and differences in biological processes between denervated skin and muscle, and between Crohn's disease and ulcerative colitis (UC) colon tissues.

Main Methods:

• Development of the categoryCompare methodology for comparing high-throughput omics data based on biological annotations.
• Application of categoryCompare to hypothetical datasets to assess its performance relative to gene-level comparisons.
• Analysis of two biological case studies: denervated skin vs. denervated muscle, and Crohn's disease vs. UC colon tissues.

Main Results:

• Annotation-level comparison using categoryCompare often yielded superior results compared to gene-level comparisons in hypothetical datasets.
• The effectiveness of annotation comparison was influenced by the number of genes within annotation terms, data noise levels, and sample combination methods.
• The skin vs. muscle denervation comparison revealed distinct tissue-specific responses, while the Crohn's vs. UC comparison highlighted shared inflammatory processes and disease-specific pathways.

Conclusions:

• CategoryCompare provides a robust framework for cross-platform and cross-sample analysis of high-throughput omics data by leveraging biological annotations.
• This annotation-centric approach overcomes limitations of feature-level comparisons, enabling the identification of conserved biological processes across heterogeneous datasets.
• The method demonstrates utility in identifying both commonalities and differences in biological responses in complex disease states and physiological conditions.