Skip to main content

Table 1 Power and Type I error rates for several simulated datasets and analysis strategies.

From: "Does replication groups scoring reduce false positive rate in SNP interaction discovery?: Response"

  

OR

Analysis Type

 

1.67

2.10

2.85

1.57 Mixed

Combined group (p = 10 -3 )

Power

90

100

100

71

 

False positive rate

0.00379

0.01144

0.02448

0.00297

3 group No Direction (p = 10 -1 )

Power

57

100

100

0

 

False positive rate

0.00197

0.00538

0.01713

0.00108

3 group Direction (p = 10 -1 )

Power

57

100

100

0

 

False positive rate

0.00121

0.00486

0.01645

0.00039

Combined group (p = 10 -4 )

Power

66

100

100

50

 

False positive rate

0.00103

0.00571

0.01824

0.00070

  1. Each dataset consists of 3000 cases and 3000 controls genotyped for 20 SNPs, and each experimental dataset was simulated 100 times. The simulated effect size for each experiment (a pair of SNPs per simulation) had an average Odds-Ratio (OR) of 1.67, 2.10, and 2.85. A final mixed model (overall OR of 1.57) resulted from the combination of two null simulations with a simulation with a pair of SNPs associated with case-status. The combined-group strategy analyzed the full sample size simultaneously (3000 cases and 3000 controls), and was applied for two different significance levels (p = 10-3 and p = 10-4). The 3-replication-group strategy analyzed three replication groups, each with 1000 cases and 1000 controls, with a significance level of p = 10-1 in each group. Results with and without applying a direction filter (consistency of results in size and direction of effect across replication groups) are shown. Power is expressed as a percentage.
Back to article page

BMC Genomics

ISSN: 1471-2164

Contact us