Table 1 Inference error for the KE DH-lines by changing a single imputing parameter
Parameterdefaultrange testedbestoverall impact
BEAGLE 5.1default (0.0127%)
ne1,000,0001 - 1,000,000100,000 (0.0125%)Figure S7
err0.0000670.01 - 0.000010.001 (0.0125%)Figure S9
window4010 - 1,000100 (0.0125%)Figure S11
burnin62 - 5050 (0.0126%)Figure S13
iterations122 - 4040 (0.0127%)Figure S15
phase-states28050 - 10,000default (0.0127%)Figure S18
imp-states, imp-segment, cluster, imp-step, imp-nsteps1,600, 6, 0.005, 0.1, 7only impacts UM imputation
BEAGLE 5.0default (0.0142%)
ne1,000,0001 - 1,000,00030,000 (0.132%)Figure S6
err0.00010.01 - 0.000010.005 (0.0141%)Figure S8
window4010 - 1,000200 (0.0140%)Figure S10
burnin62 - 50default (0.0142%)Figure S12
iterations122 - 5025 (0.0141%)Figure S14
phase-segment41 - 2510 (0.0135%)Figure S16
phase-states28050 - 1,000100 (0.0136%)Figure S17
imp-states, imp-segment, cluster, imp-step1,600, 6, 0.005, 0.1only impacts UM imputation
BEAGLE 4.1default (0.255%)
niterations50 - 25virtually no differences for DHs
modelscale0.80.5 - 51.5 (0.0438%)Figure S19
ne1,000,0001 - 1,000,00010,000 (0.254%)Figure S20
BEAGLE 4.0default (0.201%)
buildwindow1,2001 - 2,5005 (0.028%)Figure S21
singlescale0.80.5 - 51.5 (0.066%)Figure S22
nsamples41 - 5050 (0.152%)Figure S23
burnin-its52 - 2525 (0.199%)Figure S24
phase-its52 - 25virtually no differences for DHs
impute-its62 - 25only impacts UM imputation