kWIP

kWIP estimates genetic similarity between samples from sequencing data in a de novo, reference-free manner using k-mer frequencies and a weighted inner product metric.

Key Features:

• Assembly- and Alignment-Free: Operates without requiring genome assembly or read alignment, enabling reference-free analysis of sequencing data.
• k-mer-Based Analysis: Uses k-mer frequencies derived from sequencing reads as the basis for similarity estimation.
• Probabilistic Data Structure: Employs a probabilistic approach to manage and store k-mer counts efficiently for rapid computation.
• Weighted Inner Product (WIP) Metric: Calculates genetic similarity using a Weighted Inner Product metric applied to k-mer frequency data.
• Pairwise Similarity and Distance Matrix: Computes pairwise similarities between samples and produces a distance matrix for downstream analysis.
• Implementation: Implemented in C++ as stated in the source description.

Scientific Applications:

• Sample Identity Verification: Confirms whether individuals or samples match expected genetic lineages to detect mislabeling.
• Detection of Mix-Ups and Non-Obvious Variation: Identifies sample mix-ups and subtle genomic differences not apparent with some traditional methods.
• Population Structure Analysis: Aids reconstruction of true relatedness and population structure from simulated and empirical datasets.
• Re-analysis Consistency: Can reproduce relatedness patterns consistent with marker-based analyses when applied to published datasets.

Methodology:

kWIP counts k-mers from sequencing reads using a probabilistic data structure, computes pairwise similarities via a Weighted Inner Product metric on k-mer frequency data, and outputs a distance matrix; it operates without assembly or alignment.