Skip to main content

Table 2 Empirical type-I error and power obtained in 1,000 simulations using the three different approaches: REX (iterative regression), PEMM (probe-specific mixture model) and threshold.

From: Probe-specific mixed-model approach to detect copy number differences using multiplex ligation-dependent probe amplification (MLPA)

     Type-I error Power (gains) Power (loses)
σ β σ γ σ ε x ¯ a l t MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGafmiEaGNbaebadaWgaaWcbaGaemyyaeMaemiBaWMaemiDaqhabeaaaaa@31AF@ REX PEMM REX PEMM thres REX PEMM thres
0.2 0.2 0.05 3.9 0.044 0.034 0.750 0.765 0.000 0.724 0.747 0.000
0.2 0.2 0.08 3.8 0.043 0.029 0.581 0.584 0.000 0.544 0.564 0.000
0.2 0.4 0.05 4.0 0.048 0.045 0.885 0.919 0.021 0.896 0.922 0.041
0.2 0.4 0.08 3.9 0.049 0.038 0.806 0.827 0.025 0.791 0.820 0.053
0.2 0.6 0.05 4.0 0.054 0.054 0.933 0.954 0.130 0.938 0.957 0.178
0.2 0.6 0.08 4.0 0.051 0.045 0.875 0.903 0.142 0.870 0.888 0.186
0.2 1.0 0.05 3.9 0.053 0.058 0.968 0.976 0.355 0.962 0.978 0.429
0.2 1.0 0.08 4.0 0.045 0.055 0.924 0.954 0.382 0.926 0.954 0.444
0.2 1.5 0.05 4.0 0.047 0.061 0.974 0.983 0.526 0.974 0.986 0.597
0.2 1.5 0.08 4.0 0.052 0.059 0.952 0.971 0.580 0.960 0.974 0.664
0.2 4.0 0.05 3.9 0.053 0.065 0.991 0.993 0.826 0.993 0.997 0.860
0.2 4.0 0.08 4.0 0.049 0.063 0.982 0.990 0.834 0.989 0.993 0.868
0.4 0.2 0.05 4.0 0.044 0.035 0.721 0.754 0.000 0.726 0.752 0.000
0.4 0.2 0.08 3.9 0.045 0.030 0.568 0.544 0.000 0.528 0.550 0.001
0.4 0.4 0.05 3.9 0.048 0.044 0.898 0.912 0.023 0.884 0.919 0.040
0.4 0.4 0.08 4.2 0.050 0.040 0.810 0.823 0.041 0.799 0.830 0.062
0.4 0.6 0.05 4.0 0.047 0.050 0.929 0.945 0.142 0.915 0.940 0.176
0.4 0.6 0.08 4.0 0.046 0.046 0.869 0.895 0.152 0.859 0.894 0.175
0.4 1.0 0.05 4.0 0.051 0.059 0.968 0.983 0.362 0.961 0.976 0.428
0.4 1.0 0.08 3.9 0.050 0.052 0.937 0.956 0.380 0.937 0.951 0.470
0.4 1.5 0.05 4.0 0.049 0.062 0.976 0.981 0.553 0.967 0.978 0.599
0.4 1.5 0.08 4.1 0.050 0.056 0.951 0.969 0.574 0.955 0.971 0.620
0.4 4.0 0.05 3.9 0.047 0.065 0.991 0.995 0.835 0.991 0.997 0.848
0.4 4.0 0.08 4.0 0.050 0.064 0.983 0.988 0.832 0.990 0.997 0.863
0.6 0.2 0.05 4.0 0.045 0.037 0.720 0.748 0.001 0.740 0.775 0.004
0.6 0.2 0.08 4.0 0.044 0.029 0.565 0.584 0.001 0.533 0.572 0.003
0.6 0.4 0.05 4.0 0.051 0.047 0.880 0.896 0.034 0.892 0.915 0.055
0.6 0.4 0.08 4.0 0.047 0.041 0.800 0.824 0.060 0.792 0.818 0.079
0.6 0.6 0.05 4.1 0.051 0.054 0.919 0.938 0.139 0.926 0.947 0.186
0.6 0.6 0.08 3.9 0.047 0.046 0.877 0.893 0.179 0.889 0.924 0.230
0.6 1.0 0.05 4.0 0.050 0.060 0.959 0.976 0.370 0.956 0.971 0.435
0.6 1.0 0.08 4.0 0.048 0.054 0.930 0.947 0.445 0.922 0.948 0.452
0.6 1.5 0.05 4.0 0.049 0.063 0.975 0.983 0.549 0.977 0.985 0.601
0.6 1.5 0.08 3.9 0.049 0.059 0.962 0.973 0.588 0.960 0.965 0.652
0.6 4.0 0.05 3.9 0.053 0.065 0.988 0.993 0.814 0.992 0.994 0.845
0.6 4.0 0.08 3.9 0.055 0.066 0.979 0.987 0.835 0.984 0.995 0.859
  1. These results are for the case of having 10% of probes as a internal control probes (needed for the REX approach) and 10% of probes as a true altered probes. The results are given for different scenarios between probe variability (σ β ), probe-test variability (σ γ ) and within-probe variability (σ ε ). The column x ¯ a l t MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGafmiEaGNbaebadaWgaaWcbaGaemyyaeMaemiBaWMaemiDaqhabeaaaaa@31AF@ indicates the mean number of simulated altered probes.