GREAT does not return regions or gene lists

[fwzhao]

Enrichment results of GREAT do not return the matching regions or genes. I will follow up on this, perhaps changed with versioning.

[fwzhao]

I've added a new branch with additions to retrieve associated genes and regions.

One thing you might want to modify is the statistical threshold for retrieving features, otherwise it can take a really long time. So far, I have it set on p_adjusted or p_adjusted_hyper (one of the two has be significant for genes to be retrieved. And I impose a threshold of anything <= 0.1 gets annotated. This can feel too lenient / strict depending on your perspective and you could consider making this parameterizable.

It's a bad idea to annotate all terms with features... I tested this and it takes FOREVER.

rGREAT_annotate_features

[sreichl]

GREAT (pun intended), my first thought would be to re-use the p-adjusted threshold from the config for that, so that no ew config is required.

enrichment_analysis/config/config.yaml

Line 135 in 6be93bc

GREAT: 0.01

[fwzhao]

I think that's a GREAT idea ;-)

So in my experience, the best enrichments you get out of GREAT are a combination of p_adjust (strict, <=0.05) and p_adjust_hyper (less strict, <= 0.1). The former is for strong region set enrichments, while the latter accounts for where the enriched regions correspond to very few genes (e.g. 1 gene, 15 regions) - which can still be meaningful.

For data exploration, it's helpful to leave this a bit more lenient (e.g. thresholding at 0.1, and 0.2 respectively). In case you might want to consider this in the "defaults" (not really, I know) specified in the config file.

[sreichl]

'a' %in% background TRUE
also in gene sets! check input txt -> it's from the GREAT region to gene mapping. Why? Hopefully resolved once the mapping is provided differently