CARE

CARE performs scalable alignment-based error correction of Illumina next-generation sequencing (NGS) reads to reduce false-positive corrections and improve data quality for downstream analyses such as de novo genome assembly.

Key Features:

• Alignment-Based Approach: CARE leverages alignment-based strategies to correct sequencing errors by inspecting multiple alignments rather than relying on independent k-mer analysis.
• Minhashing Technique: CARE uses minhashing to perform similarity searches across large read collections, enabling rapid identification of related reads for high-quality multiple alignments.
• Scalability and Performance: CARE processes large datasets, demonstrated by handling a human-genome Illumina NGS dataset in approximately 4 hours on a single workstation, and supports GPU acceleration for enhanced performance.
• Error Correction Accuracy: CARE generates significantly fewer false-positive corrections while maintaining competitive true-positive rates compared with Musket, SGA, BFC, Lighter, Bcool, and Karect.
• Application in Genome Assembly: When applied prior to assembly in NGS pipelines, CARE improves de novo assembly results across multiple real datasets.

Scientific Applications:

• Genome Sequencing Error Correction: CARE improves base-level accuracy of Illumina NGS reads used in genome sequencing projects.
• De novo Genome Assembly: CARE's reduction of false positives and preservation of true positives enhances contiguity and correctness of de novo assemblies.
• Large-scale and Complex Genomes: CARE's scalability and accuracy benefit sequencing projects involving complex organisms or large-scale datasets.

Methodology:

CARE performs minhashing-based similarity searches to identify related reads, computes and inspects high-quality multiple alignments for alignment-based error correction instead of independent k-mer fragmentation, and is implemented in C++ with an optional CUDA/C++ GPU-accelerated version.