Correction factors help minimize the number false positive results when making statistical predictions based on large data sets. Below are some of the key correction factors that are available in iPathwayGuide.
False Discovery Rate
The False Discovery Rate (FDR) is designed to control the rate of false positives in the set of rejected null hypotheses. This method accepts that some of the results will be false positives. Given a set of rejected null hypotheses at 5%, FDR guarantees you will have less than 5% false positives in the set.
Bonferonni
Unlike FDR, Bonferonni focuses on eliminating all the false positives, and therefore is one of the most stringent correction methods. Given a set of rejected null hypotheses at 5%, Bonferonni guarantees there is a less than a 5% chance that th set contains any false positives.
Elim Pruning
The Elim method assesses the ontology terms from the bottom up depending on significance. Before computing the significance for a parent term, all the genes annotated to significant child terms are "eliminated" from the parent. Hence the name.
The benefit of this approach is that specialized terms become more important than generalized terms (e.g. "response to amphetamine" is more descriptive than "response to chemical”).
Weight Pruning
Unlike elim, the Weight method does not eliminate the genes from the parent. Instead, it assigns a weight to each gene depending the significance of the neighboring terms containing that gene.
This weighted approach is less stringent than elim - capturing more true positives with the drawback of including additional false positives.
iPathwayGuide is the only tool to provide these advanced correction factors to help you minimize false positives. Try it today for free and see what is truly significant in your data.