This mode is used to perform univariate feature selection with a permutation Test. We compare each feature individually for evidence of differential distribution in the two prediction classes. If the adhoc test statistic p-value is less than the confidence interval, then the feature is selected as an informative biomarker.
It is implemented by org.bdval.DiscoverWithPermutation.java.
Mode Parameters
The following options are available in this mode
| Flag | Arguments | Required | Description |
|---|---|---|---|
(-n|--alpha) |
alpha | yes | The significance level for the permutation test p-value. Default 0.05/5% confidence level. (default: 0.05) |
(-z|--numberOfPermutations) |
numberOfPermutations | yes | The number of permutation iterations. Default 1000. (default: 1000) |
--output-gene-list |
n/a | no | Write features to the output in the tissueinfo gene list format. |
--report-max-probes |
report-max-probes | no | Restrict output to the top ranked probes. This option works in conjunction with alpha and can further restrict the output. |
Leave a Comment