ntStat: k-mer characterization using occurrence statistics in raw sequencing data

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• K-mer counts are essential for genomic analysis tasks like genome assembly and variant detection.
• Existing k-mer counting tools use various methods for efficiency, including parallelism and probabilistic data structures.
• K-mer count distributions reveal crucial genomic features such as genome size, heterozygosity, and basecalling quality.

Purpose of the Study:

• To introduce ntStat, a novel toolkit for k-mer counting and depth analysis.
• To leverage succinct Bloom filter data structures for efficient tracking of k-mer information.
• To enable de novo inference of genomic insights from k-mer data.

Main Methods:

• Utilized succinct Bloom filter data structures for tracking k-mer counts and depth.
• Employed evolutionary computation to model k-mer count histograms.
• Compared ntStat's performance against established tools like DSK, BFCounter, Jellyfish, and KMC3.

Main Results:

• ntStat demonstrated superior speed compared to DSK, BFCounter, hackgap, and Squeakr.
• ntStat used less memory than other non-disk-based k-mer counters, achieving 99.5-99.9% accuracy.
• The histogram analysis module accurately estimated heterozygosity and k-mer coverage in simulated long-read datasets.

Conclusions:

• ntStat offers an efficient and accurate solution for k-mer counting and genomic analysis.
• The toolkit provides valuable de novo insights into genome characteristics and sequencing data quality.
• ntStat represents a significant advancement in k-mer analysis tools for large-scale genomics research.