Framework for Parallel Preprocessing of Microarray Data Using Hadoop.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Microarray technology is crucial for gene expression studies and disease diagnosis.
• Public biological data from NCBI requires preprocessing to remove noise and bias.
• Robust Multiarray Average (RMA) is a standard but time-consuming preprocessing method for large datasets.

Purpose of the Study:

• To parallelize the RMA algorithm for efficient microarray data preprocessing.
• To leverage Hadoop's distributed computing and R's statistical capabilities.
• To address the computational challenges of processing large-scale microarray datasets.

Main Methods:

• Parallelization of the RMA algorithm using Hadoop and R.
• Implementation on a 5-node cluster with 16 cores and 16GB memory per node.
• Comparison of the proposed parallelized RMA with sequential RMA and affyPara.

Main Results:

• The parallelized RMA using Hadoop and R demonstrated superior performance.
• Significant speed-up rates were achieved compared to sequential and affyPara methods.
• The proposed approach efficiently handles large volumes of microarray data.

Conclusions:

• Parallelizing RMA with Hadoop and R offers an efficient solution for microarray data preprocessing.
• This approach overcomes the limitations of traditional methods for large datasets.
• The findings have implications for accelerating genomic data analysis.