IndGOterm: a qualitative method for the identification of individually dysregulated GO terms in cancer

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Individual pathway analysis is crucial for understanding cancer patient heterogeneity and guiding personalized therapy.
• Batch effects from diverse data technologies limit the application of many existing individual pathway analysis methods.
• Relative expression ordering (REO) based methods show promise in mitigating batch effects.

Purpose of the Study:

• To propose an individual qualitative Gene Ontology (GO) term analysis method, IndGOterm, based on REO.
• To evaluate IndGOterm's performance against established single-sample enrichment analysis methods like ssGSEA and GSVA.
• To demonstrate IndGOterm's utility in capturing cancer-relevant biological information and revealing patient-specific characteristics.

Main Methods:

• Developed IndGOterm, an individual qualitative GO term analysis method utilizing gene REO.
• Compared IndGOterm with ssGSEA and GSVA in terms of batch effect insensitivity.
• Applied IndGOterm to survival and drug response data to assess its ability to identify cancer-associated terms.

Main Results:

• IndGOterm effectively ignores batch effects, outperforming ssGSEA and GSVA in this regard.
• IndGOterm identified more cancer-related terms compared to other methods in survival and drug response analyses.
• The method revealed distinct dysregulation patterns within homologous patients, highlighting cancer's heterogeneity.

Conclusions:

• IndGOterm is a robust tool for individual pathway analysis, insensitive to batch effects.
• It enhances the capture of clinically relevant biological information from cancer patient data.
• IndGOterm facilitates the discovery of intrinsic cancer characteristics and patient-specific dysregulation models.