When a discrete variable is defined, a differential analysis can be performed. Samples with the same value of the discrete variable are pooled into groups. Each probe is then compared using Anova analysis across the groups. The Anova (analysis of variation) seeks to identify whether the defined groups shows statistically significant differences in expression of the probe. The p-value for each probe is computed by looking at a t-test of the null hypothesis that the variation of the probe between groups could have arisen by chance. The inter-group variance is compared to the intra-group variance to derive a t-statistic for each probe.
After all probes have been subjected to Anova analysis, the probes are ranked in order of the most statistically significant change.
When using a multiplex assay, the issue of multiple
comparisons should be kept in mind. A statistical procedure used to identify significance for a probe tested in isolation is likely to generate many "significant" results by sheer chance when many probes are tested at once, just as the odds of winning the lottery with 70 tickets are better than with 1. Thus using unadjusted p-values is likely to lead to false positives, especially in the context of a large PLX file.
To compensate for the statistical likelihood of finding an apparently significant result when sufficiently many different probes are analyzed, a "raw" and an "adjusted" score is presented. The raw score defines the statistical significance of the differential expression as if that probe had been the only probe. The adjusted score corrects for multiple
comparisons using the Bonferroni method
The Anova results are presented in a table and a chart on the Groups tab.
In addition, a "volcano" plot of the significance relative to the fold change summarizes the data.
To access differential analysis
To sort the probes in order of significant difference of
- Select the heatmap tab and under the probe controls, select the desired variable as the "analysis" option.
- Click "Use probe order" button at the bottom of the heatmap controls
You may need to reselect the variable in the "analysis" pull-down.
- Select the Groups tab to see the table and chart
Help Front Page
Anova analysis (Wikipedia)
Multiple comparisons (Wikipedia)
Bonferroni method (Wikipedia)
Volcano Plot (Wikipedia)