Clique-Based Clustering of Correlated SNPs in a Gene Can Improve Performance of Gene-Based Multi-Bin Linear Combination Test
Table 3
The average over 1000 genes of the number of clusters per gene, the mean size of the clusters within a gene, and the standard deviation of the cluster sizes within a gene for two clustering methods (LDSelect and CLQ).
Allele frequency cut
# of clusters*
Mean size of clusters*
SD size of clusters*
LDSelect
CLQ
LDSelect
CLQ
LDSelect
CLQ
.05
.3
1.84
2.94
6.39
3.70
2.55
2.60
.4
2.43
3.27
4.82
3.40
2.82
2.33
.5
3.02
3.67
3.85
3.06
2.49
2.02
.6
3.65
4.19
3.13
2.67
2.13
1.75
.7
4.36
4.79
2.61
2.32
1.75
1.46
.8
5.18
5.52
2.18
2.01
1.38
1.19
.9
6.29
6.57
1.76
1.65
1.00
0.89
.01
.3
2.40
3.53
5.25
3.28
3.33
2.53
.4
3.02
3.91
4.08
3.00
3.02
2.25
.5
3.66
4.36
3.32
2.71
2.52
1.96
.6
4.35
4.90
2.75
2.40
2.07
1.67
.7
5.10
5.57
2.34
2.10
1.66
1.40
.8
5.99
6.32
1.98
1.85
1.31
1.15
.9
7.17
7.42
1.64
1.56
0.94
0.85
The differences of the obtained characteristics within genes are compared by paired -test and all results were significant with values <1 except the italic pairs ().
