Rapid, Phase-free Detection of Long Identity-by-Descent Segments Enables Effective Relationship Classification

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Identity-by-descent (IBD) segments are crucial for genetic analyses like demographic inference and relationship classification.
• Current IBD detection methods often require computationally intensive phasing, limiting scalability with growing genetic datasets.

Purpose of the Study:

• To develop and evaluate IBIS, a novel algorithm for rapid and accurate detection of IBD segments in unphased genetic data.
• To benchmark IBIS against existing IBD detection tools in terms of speed and accuracy.

Main Methods:

• Developed IBIS, an algorithm for identifying long regions of allele sharing between unphased individuals.
• Benchmarked IBIS against Refined IBD, GERMLINE, and TRUFFLE using simulated and real genetic datasets.
• Assessed performance based on runtime, accuracy of IBD segment inference (≥7 cM), and classification of relative degrees.

Main Results:

• IBIS achieved significant speedups (805-946×) compared to phasing followed by Refined IBD or GERMLINE, completing in minutes versus days.
• IBIS demonstrated comparable accuracy to Refined IBD and GERMLINE for inferring IBD segments.
• IBIS accurately classified relatives and showed robustness in admixed populations, outperforming allele frequency-based methods like KING.

Conclusions:

• IBIS offers a highly scalable and accurate solution for IBD segment detection, addressing the computational challenges of large genetic datasets.
• The method provides a valuable tool for genetic relationship inference and demographic studies, particularly in diverse populations.