Software

Parallel computing for efficient inbreeding coefficient calculation

---

Abstract

In this study, we evaluated the parallel scalability of two approaches for calculating inbreeding coefficients. While multi-threaded parallel computing is routinely used in genetic evaluations to accelerate computations, its application to inbreeding coefficient calculation has not been presented. Several algorithms have been presented for computing inbreeding coefficients. In particular, Sargolzaei and Iwaisaki compared four such algorithms. We modified the Sargolzaei and Iwaisaki (SI) algorithm to use parallel computing by OpenMP in Fortran. We also implemented an OpenMP parallel version of the recursive tabular method to illustrate efficiency differences already reported in their study. We evaluated both parallel implementations using an Irish beef cattle pedigree having 9.5 million individuals. For the tabular (SI in parenthesis) approach, the use of 2, 5, 10, and 15 CPU threads gave speed-ups of 2.0 (1.9), 4.7 (4.3), 8.9 (8.1), and 14.2 (13.0), respectively, when compared to the single thread computations. Correspondingly, peak memory use relative to the single thread program increased 1.8 (4.0), 3.1 (7.3), 4.8 (12.9), and 6.9 (18.6) times for the tabular (SI) approach. In the single-thread run, the tabular approach required 31.9 hours, while the SI algorithm required only 1.1 minutes, and the peak memory use was 176 MB for the tabular approach, but 503 MB for the SI algorithm. With 15 CPU threads, the SI method needed only 5 seconds and used 3.2 GB of memory, when the tabular method took over 2 hours and used 1.2 GB of memory. In conclusion, both algorithms were efficiently parallelized and showed similar speed-ups by increased thread number. However, the parallel SI version had a higher memory use than the tabular method. Nevertheless, the SI approach showed superior computational efficiency over the tabular method even with parallel computing.

Keywords: 2026

How to Cite:

Strandén, I., Evans, R. & Taskinen, M., (2026) “Parallel computing for efficient inbreeding coefficient calculation”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2283178. doi: https://doi.org/10.31274/wcgalp.23491

Rights: 1