Associating Multivariate Traits with Genetic Variants Using Collapsing and Kernel Methods with Pedigree- or Population-Based Studies
Table 2
Empirical type I errors of the five competing methods with binary traits.
Marker-specific weight
Nominal level
Working correlation
Method
HoK3
HoO
HeK
HeO
BT
Unweighted marker-specific weight1
0.05
U/U2
0.04944
0.05154
0.05086
0.05280
0.04952
E/E
0.04930
0.05144
0.05068
0.05318
0.04946
0.01
U/U
0.00974
0.00994
0.00982
0.01026
0.01000
E/E
0.00974
0.00998
0.00984
0.01028
0.00998
0.001
U/U
0.00068
0.00084
0.00100
0.00098
0.00106
E/E
0.00066
0.00084
0.00102
0.00094
0.00104
0.0001
U/U
0.00008
0.00002
0.00012
0.00010
0.00000
E/E
0.00008
0.00002
0.00012
0.00010
0.00002
Weighted marker-specific weight
0.05
U/U
0.05170
0.04900
0.05256
0.04922
0.04576
E/E
0.05168
0.04920
0.05232
0.04930
0.04556
0.01
U/U
0.01028
0.00976
0.00996
0.00972
0.00886
E/E
0.01024
0.00982
0.00986
0.00976
0.00884
0.001
U/U
0.00110
0.00080
0.00096
0.00090
0.00088
E/E
0.00112
0.00076
0.00096
0.00088
0.00090
0.0001
U/U
0.00004
0.00008
0.00010
0.00012
0.00006
E/E
0.00006
0.00008
0.00010
0.00012
0.00008
1The unweighted marker-specific weight is given by ; the weighted marker-specific weight is given by .2U/U represents the structures of the working within-cluster and multivariate-response correlation matrices considered by the unstructured structures; E/E represents the structures of the working within-cluster and multivariate-response correlation matrices considered by the exchangeable structures. 3HoK, HoO, HeK, HeO, and BT are our proposed methods.