Kmer2SNP

Kmer2SNP calls single nucleotide polymorphisms (SNPs) directly from raw next-generation sequencing reads without requiring a reference genome or sequence alignment.

Key Features:

• Reference-Free Approach: Operates without a reference genome to enable SNP calling in species lacking high-quality references.
• Alignment-Free Process: Detects SNPs without performing sequence alignment.
• K-mer Frequency Analysis: Computes k-mer frequency distributions from raw reads to identify potential heterozygous k-mers that appear in only one haplotype.
• Graph Construction: Represents heterozygous k-mers as graph vertices and connects k-mers likely corresponding to SNPs with edges.
• Edge Weighting: Assigns weights to edges based on overlapping information between the connected k-mers.
• Maximum Weight Matching: Computes a maximum weight matching on the k-mer graph to select SNP candidate edges representing variant calls.
• Benchmarking: Benchmarked against hybrid (assembly-based) and assembly-free reference-free methods on simulated and real datasets, reporting superior SNP calling quality and an order-of-magnitude faster performance.

Scientific Applications:

• Gene-Disease Diagnosis: Facilitates identification of disease-associated variants.
• Drug Design: Supports analysis of genetic variation relevant to drug response.
• Ancestry Inference: Supports population genetics and evolutionary biology studies.

Methodology:

Computes k-mer frequency distributions from raw reads to detect heterozygous k-mers, constructs a graph with heterozygous k-mers as vertices and edges connecting k-mer pairs likely corresponding to SNPs, assigns edge weights based on k-mer overlap information, and computes a maximum weight matching to output SNP candidates as edges between k-mer pairs.