Efficient Implementation of the AI-REML Iteration for Variance Component QTL Analysis
Regions in the genome that affect complex traits, quantitative trait loci (QTL), can be identified using statistical analysis of genetic and phenotypic data. When restricted maximum-likelihood (REML) models are used, the mapping procedure is normally computationally demanding. We develop a new efficient computational scheme for QTL mapping using variance component analysis and the AI-REML algorithm. The algorithm uses an exact or approximative low-rank representation of the identity-by-descent matrix, which combined with the Woodbury formula for matrix inversion results in that the computations in the AI-REML iteration body can be performed more efficiently. For cases where an exact low-rank representation of the IBD matrix is available a-priori, the improved AI-REML algorithm normally runs almost twice as fast compared to the standard version. When an exact low-rank representation is not available, a truncated spectral decomposition is used to determine a low-rank approximation. We show that also in this case, the computational efficiency of the AI-REML scheme can often be significantly improved.
💡 Research Summary
The paper addresses the computational bottleneck inherent in variance‑component quantitative trait locus (QTL) mapping when restricted maximum‑likelihood (REML) estimation is employed. In the standard AI‑REML (average information REML) algorithm, each iteration requires the evaluation of the covariance matrix Σ = σ_g²K + σ_e²I and its inverse, where K is the identity‑by‑descent (IBD) matrix describing genetic relationships among N individuals. Because K is dense and of size N × N, naïve computation of Σ⁻¹ incurs O(N³) operations, making genome‑wide analyses with thousands or tens of thousands of samples impractical.
The authors propose a two‑pronged strategy that dramatically reduces this cost by exploiting low‑rank structure in K and applying the Woodbury matrix identity.
- Exact low‑rank representation – In many breeding designs the IBD matrix can be expressed exactly as K = U Uᵀ, where U is N × r with r ≪ N (the true rank of K or the number of independent genetic components). Substituting this form into Σ yields Σ = σ_e²I + σ_g²U Uᵀ. The Woodbury formula gives
Σ⁻¹ = σ_e⁻²
Comments & Academic Discussion
Loading comments...
Leave a Comment