Table 3.
Type I error rates of the different approaches for the small sample scenarios.
Scenario | CAST | PC3CV | LMMCV | LocPerm |
---|---|---|---|---|
50SE-100SE | 0.0012 | 0.0015 | 0.0012 | 0.0009 |
50SE-1000E | 0.0016 | 0.0012 | 0.0028 | 0.0008 |
50SE-1000W | 0.0046 | 0.0011 | 0.0015 | 0.0010 |
50SE-2000W | 0.0046 | 0.0010 | 0.0016 | 0.0011 |
50E-100E | 0.0014 | 0.0015 | 0.0012 | 0.0010 |
50E-1000E | 0.0010 | 0.0010 | 0.0021 | 0.0009 |
50E-1000W | 0.0051 | 0.001 | 0.0014 | 0.0010 |
50E-2000W | 0.0050 | 0.0009 | 0.0015 | 0.0011 |
50World-100W | 0.0013 | 0.0015 | 0.0012 | 0.0010 |
50World-1000E | 0.0077 | 0.0007 | 0.0053 | 0.0010 |
50World-1000W | 0.0009 | 0.0010 | 0.0021 | 0.0009 |
50World-2000W | 0.0009 | 0.0009 | 0.002 | 0.0010 |
The nominal level alpha considered is . Type I error rates under the lower bound of the 95%PI are displayed in italic and above the upper bound of the 95%PI in bold.
Supplementary Table S3 provides the adjusted 95%PI for the different number of genes tested in each scenario.