@@ -418,9 +418,9 @@ which account for the top `64` obesity transcripts. We note that some of these a
# Example 8: Latent data assumptions. Building a Parent-Child list
So you are sitting on a large amount of groupings that you have a significance test for what you are interested in. Since you will conduct a large amount of tests there is also a large risk that you will technically test the same thing over and over again. In order to remove this effect from your group testing you could employ my `HierarchicalEnrichment` routine, but then you will need a relationship file describing how to build the group DAG Hierarchy. The relationship file contains a parent id, a tab delimiter and a child id on each line. The routine that I demonstrate here uses a divide-and-conquer type approach which means that a subgroup, or child, is only assigned if it is fully contained within the parents definition. You can create redundant assignments by setting `bSingleDescent=False` but it is not the recommended default setting.
So you are sitting on a large amount of groupings that you have a significance test for. Testing what you are interested in per analyte symbol/id. Since you will conduct a large amount of tests there is also a large risk that you will technically test the same thing over and over again. In order to remove this effect from your group testing you could employ my `HierarchicalEnrichment` routine, but then you will need a relationship file describing how to build the group DAG Hierarchy. The relationship file contains a `parent id`, a `tab delimiter` and a `child id` on each line. The routine that I demonstrate here uses a divide-and-conquer type approach which means that a subgroup, or child, is only assigned if it is fully contained within the parents definition. You can create redundant assignments by setting `bSingleDescent=False` but it is not the recommended default setting.
Have no fear! This can be done with my `build_pclist_word_hierarchy` routine. Enough talk. Let us assume that you are sitting on the following data:
Construction of the downward node relationships can be done with my `build_pclist_word_hierarchy` routine. Ok, Enough talk. Let us assume that you are sitting on the following data:
If there are latent assumptions for some groupings then you can also read them out by checking what the definition referes to (here we already know that there is one for the mitochondrion definition):
That the definition for the mitochondrion is fully contained within the melanosome membrane definition and so testing for that group should be accounted for when testing the parent.
That the definition for the mitochondrion is fully contained within the melanosome membrane definition and so testing for that group should be accounted for when testing the parent. We know that the melanosome membrane is associated with sight and that being diabetic is associated with mitochondrial dysfunction, but also that diabetic retinopathy affects diabetics, so there might be a knowledge based, or True, genetic connection relating these two spatially distinct regions of the cell.
