Skip to content

UserGuide_MutualExclusivity.rst

Latest commit

 

History

History
62 lines (42 loc) · 2.81 KB

UserGuide_MutualExclusivity.rst

File metadata and controls

62 lines (42 loc) · 2.81 KB

Mutual exclusion and co-occurrence test

Contents

There are two scenarios where Gitools can perform a mutex or co-oc test:

  1. When sorting heatmap rows according to mutual exclusivity of :doc:`UserGuide_DataEvents` it is possible to also carry out a test of significance for the mutual-exclusive or co-occurring distribution of those events
  2. Having row and/or column annotation which describe subgroups of each dimension allow to carry out a test for each combination of column-row group via the Analysis - menu

Data files needed

Data Matrix: A file with either binary (i.e. presence/absence), categorical (i.e. alteration status) or continuous (i.e. an expression matrix) data.

Running the analysis

  1. Open the data matrix as heatmap
  2. Make sure that your data is shown with an accurate color scale.
  3. Select Edit->Rows->Sort by mutual exclusion.
  4. Add the id's of the items you want to sort and select perform statistical test

Things to consider

The test

The test is based on weighted permutations assessing the deviation of the observed coverage (number of columns with a signal) compared to expected obtained by permuting events, maintaining the number of events per row and weighted permutations for columns.

Data used

The column weights, used for the permutations are based on the "events" (:doc:`UserGuide_DataEvents`) in each column including hidden rows. The more rows that are included, the more accurate the weight parameter, therefore if you are performing the test on a dataset that contains only is a subset and rows are missing (e.g. not all genes present) this parameter may be inaccurate and the end result may be inaccurate.

The mutual exclusive result

The result of a test will yield the following values:

Statistics

  • Z-score: Zscore shows the deviation of the observed coverage (number of columns with a signal) compared to expected, obtained by permuting events, maintaining the number of events per row and weighted permutations for columns.
  • MutEx p-value: Significance p-value of mutual exclusivity derived from the Z-score
  • Co-occurrence p-value: Significance p-value of co-occurrence derived from the Z-Score
Other measures
  • Signal: Number of positive events (see :doc:`UserGuide_DataEvents`) within the data selected data.
  • Coverage: Number of columns with at least a signal of one.
  • Sig/Cov Ratio: Ratio of Signal to Coverage
  • Mean coverage: Mean coverage of the 10'000 permutations
  • Variance: The variance of the coverage from the 10'000 permutations