ParDRe: faster parallel duplicated reads removal tool for sequencing studies.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Next-generation sequencing (NGS) technologies frequently produce redundant reads.
• These duplicated reads consume significant computational resources without adding biological value.
• Efficient removal of these reads is crucial for optimizing downstream genomic analyses.

Purpose of the Study:

• To develop ParDRe, a novel de novo parallel tool for identifying and removing duplicated and near-duplicated sequencing reads.
• To enhance the efficiency of bioinformatics pipelines by reducing memory and computational demands.

Main Methods:

• ParDRe utilizes a clustering approach for Single-End or Paired-End sequences in fasta or fastq files.
• A unique bitwise method is employed for comparing DNA string suffixes.
• Hybrid Message Passing Interface (MPI) and multithreading are implemented for parallel processing on multicore systems.

Main Results:

• ParDRe effectively removes duplicated and near-duplicated reads.
• The tool demonstrates significant speed improvements, achieving up to 27.29 times faster performance than the state-of-the-art tool Fulcrum.
• Performance gains were observed on a multicore system with two 8-core processors.

Conclusions:

• ParDRe offers a highly efficient solution for managing redundant sequencing data.
• The tool's parallel architecture and novel comparison method lead to substantial reductions in runtime and resource utilization.
• ParDRe is a valuable asset for accelerating and optimizing next-generation sequencing data analysis.