ADaCGH2: parallelized analysis of (big) CNA data.

Area of Science:

• Genomics
• Bioinformatics

Background:

• Genomic DNA copy number alteration studies involve massive datasets with millions of probes and thousands of subjects.
• Current software solutions face significant performance limitations due to slow processing speeds and high memory demands.

Purpose of the Study:

• To develop a computationally efficient R package for analyzing large-scale genomic DNA copy number alteration data.
• To overcome the memory and speed limitations of existing bioinformatics tools.

Main Methods:

• Developed ADaCGH2, a BioConductor package implementing parallelized segmentation algorithms.
• Utilized multicore computing (forking) and cluster computing (MPI) for parallelization.
• Employed ff objects for efficient data reading and storage to handle large datasets.

Main Results:

• ADaCGH2 successfully analyzes datasets with up to 6 million probes per array, even those exceeding available memory.
• Achieved significant speedups of 25-40 times compared to non-parallelized versions on a 64-core machine.
• Demonstrated the feasibility of analyzing previously intractable large-scale genomic datasets.

Conclusions:

• ADaCGH2 provides a powerful and efficient solution for analyzing large genomic DNA copy number alteration datasets.
• The package enhances the feasibility and speed of complex genomic research.
• ADaCGH2 is available as an R package from BioConductor.