High dimensional surrogacy: computational aspects of an upscaled analysis

Area of Science:

• Computational Biology
• Bioinformatics
• Drug Discovery

Background:

• Genomic biomarker identification is crucial for drug discovery, often involving high-dimensional, multi-source datasets.
• Integrating these diverse datasets presents a significant computational challenge.
• High-Performance Computing (HPC) is essential for analyzing large-scale biological data.

Purpose of the Study:

• To propose a novel data analysis framework for utilizing R in a computer cluster environment.
• To address the specific challenges of multi-source data integration in high-dimensional drug discovery datasets.
• To optimize the analysis of genomic and pathway data for new drug development.

Main Methods:

• Development of a new data analysis workflow designed for R on computer clusters.
• Application of the proposed framework to a real-world, high-dimensional drug discovery dataset.
• Comparative analysis against existing R packages for parallel computing.

Main Results:

• The proposed framework demonstrates effective data analysis for high-dimensional, multi-source drug discovery data.
• The workflow is optimized for the specific data structures encountered in drug development.
• Performance comparison indicates advantages over general-purpose R parallel computing packages.

Conclusions:

• The developed R framework provides an optimized solution for HPC in drug discovery data analysis.
• This approach facilitates more efficient identification of genomic biomarkers and related biological pathways.
• The study highlights the need for specialized HPC solutions tailored to the complexities of drug discovery data.