Multiomic Integration of Public Oncology Databases in Bioconductor

Area of Science:

• Bioinformatics
• Cancer Genomics
• Computational Biology

Background:

• Cancer research increasingly relies on multiomic data from complementary genomic assays.
• Managing and analyzing complex multiomic datasets, including those from The Cancer Genome Atlas (TCGA) and cBioPortal, presents significant challenges.
• Existing computational methods struggle with integrating diverse data types from these resources, leading to errors and inefficiencies.

Purpose of the Study:

• To develop a novel and powerful approach for creating fully integrated representations of multiomic, pan-cancer databases.
• To provide R/Bioconductor packages that facilitate the management and analysis of TCGA and cBioPortal data.
• To simplify the integration of diverse experimental assays with clinicopathological data for cancer research.

Main Methods:

• Developed the curatedTCGAData and cBioPortalData R/Bioconductor packages.
• Utilized the MultiAssayExperiment data structure for integrating diverse multiomic datasets.
• Implemented out-of-memory data representation for large methylation datasets to ensure responsive loading times and analysis on memory-limited machines.

Main Results:

• Created integrated multiomic datasets from TCGA and cBioPortal using R/Bioconductor packages.
• Demonstrated simplified multiomic and pan-cancer analyses through the developed suite of tools.
• Enabled coordination of diverse experimental assays with clinicopathological data, reducing data management burden.

Conclusions:

• The developed R/Bioconductor packages provide integrated representations of multiomic cancer data.
• These tools enable analysts and developers to apply statistical and plotting methods to extensive multiomic data with user-friendly commands.
• Facilitated greatly simplified multiomic and pan-cancer analyses, advancing cancer genomics research.