Table 2 Properties of the 12 relationship inference methods we analyzed
MethodVersionCitationTypeOutputParallelized?Runtime (× cores if > 1) [× number of runs]Requires independent markersInput required from outside programAccounts for population structure
ERSA2.0Li et al. (2014)IBD segment-basedDegree of relatednessN14.3 + 96.3 hr (×16)aNIBD segmentsNA
fastIBDBeagle 3.3.2Browning and Browning (2011a)IBD segment-findingIBD segmentsN55.2 hr [× 10]NNANA
GERMLINE (-haploid)1.5.1Gusev et al. (2009)IBD segment-finding (distinguishes IBD1 and IBD2)IBD segmentsN19.2 min + 96.0 hr (×16)bNPhased genotypesNA
HaploScoreNADurand et al. (2014)IBD segment-basedIBD segmentsN2.4 + 96.3 hr (×16)aNIBD segments; phased genotypesNA
IBDseqr1206Browning and Browning (2013a)IBD segment-findingIBD segmentsY33.1 hr (×16)NNANA
KING (KING-robust)1.4Manichaikul et al. (2010)Allele frequency-based IBD estimateIBD 0,1,2 proportionsN4.6 minYNAY
PC-Relate2.0.1Conomos et al. (2016)Allele frequency-based IBD estimateIBD 0,1,2 proportionsN8.9 hr + 4.6 mincYPairwise kinship coefficientsY
PLINK 1.91.90b2kChang et al. (2015)Allele frequency-based IBD estimateIBD 0,1,2 proportionsN18.1 secYNAN
PREST-plus4.1Sun (2012)Allele frequency-based; uses linkage modelIBD 0,1,2 proportionsN178.9 hrNNAN
REAP1.2Thornton et al. (2012)Allele frequency-based IBD estimateIBD 0,1,2 proportionsN3.8 + 2.8 hrdYAncestral population allele frequencies; sample ancestry proportionsY
Refined IBDBeagle 4.1Browning and Browning (2013b)IBD segment-finding (distinguishes IBD1 and IBD2)IBD segmentsY96.0hr (× 16) [× 3]NNANA
RelateAdmix0.1Moltke and Albrechtsen (2014)Allele frequency-based IBD estimateIBD 0,1,2 proportionsY15.8 hr (×16) + 2.8 hrdYAncestral population allele frequencies; sample ancestry proportionsY
  • Type indicates the inference methodology the program uses. Runtime is wall clock time to run the program with any additional time to run programs needed for input as indicated. We ran parallelized programs using the numbers of cores indicated in parentheses, and ran fastIBD and Refined IBD multiple times as recommended by the authors, with counts indicated in square brackets. Input required from outside program indicates extraneous information needed to run the program. Programs that use either principal components, sample ancestral population proportions, or that use a model designed for multiple populations are indicated as accounting for population structure. “Y” indicates yes, “N” indicates no, and “NA” indicates not applicable. Runtimes are from a machine with four AMD Opteron 6176 2.30 GHz processors (64 cores total) and 256 GB memory.

  • a Additional time to phase the data using Beagle 4.1 and run GERMLINE.

  • b Additional time to phase the data using Beagle 4.1.

  • c Additional time to obtain KING relatedness estimates; base PC-Relate time is the sum of time to run this method and PC-AiR (Conomos et al. 2015).

  • d Additional time to obtain ancestral population proportions using ADMIXTURE (Alexander et al. 2009).