S-conLSH: alignment-free gapped mapping of noisy long reads

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single Molecule, Real-Time (SMRT) sequencing offers long reads and low GC bias, advancing genome analysis.
• Traditional alignment methods struggle with distant homologies in genomes due to genetic duplication and recombination.
• Accurate read mapping is a critical and computationally intensive step in SMRT sequencing analysis.

Purpose of the Study:

• To develop a novel, fast, and accurate alignment-free method for mapping long sequencing reads.
• To overcome the limitations of existing aligners in identifying distant homologies in complex genomes.

Main Methods:

• Introduced S-conLSH, a novel mapper utilizing Spaced context based Locality Sensitive Hashing.
• Employed multiple spaced patterns for gapped mapping of noisy long reads to reference genomes.
• Evaluated performance on five real and simulated datasets, comparing against existing methods like lordFAST.

Main Results:

• S-conLSH demonstrated at least a 2-fold increase in speed compared to lordFAST.
• Achieved 99% sensitivity on human simulated sequence data without traditional base-to-base alignment.
• Provides alignment-free mapping in PAF format by default, with an option for SAM-file output.

Conclusions:

• S-conLSH is a pioneering alignment-free tool for reference genome mapping with high sensitivity.
• The spaced-context approach effectively extracts distant similarities, crucial for complex genomic regions.
• Variable-length spaced patterns enhance flexibility for gapped mapping of noisy long reads, advancing alignment-free analysis.