A probabilistic multi-omics data matching method for detecting sample errors in integrative analysis.

Area of Science:

• Bioinformatics
• Genomics
• Data Science

Background:

• Large-scale omics data generation is prone to inevitable errors like sample swapping and mislabeling.
• Data errors can obscure true biological signals and lead to erroneous scientific conclusions.
• Accurate data annotation is critical for merging different omics data types in integrative analyses.

Purpose of the Study:

• To develop a robust method for identifying and correcting data annotation and sample errors in large omics databases.
• To ensure the integrity of multi-omics datasets before integrative analysis.

Main Methods:

• Developed proMODMatcher, a probabilistic multi-omics data matching procedure.
• Applied the procedure to simulated datasets to assess statistical power.
• Utilized proMODMatcher on The Cancer Genome Atlas and International Cancer Genome Consortium multi-omics datasets.

Main Results:

• proMODMatcher demonstrated robust statistical power, even with small cis-associations or large sample sizes.
• Identified sample errors in multiple cancer datasets from The Cancer Genome Atlas and International Cancer Genome Consortium.
• The procedure successfully identified the source of sample-labeling errors.

Conclusions:

• Sample-labeling errors are prevalent in large multi-omics datasets.
• Correction of these errors is essential prior to conducting integrative analyses.
• Ensuring data accuracy is crucial for reliable omics research.