Estimate effect size of GO over-representation

This is a crude function to estimate the effect size of GO over-representation i.e. we know a term is over-represented, but we want to estimate the effect size/how over-represented it is. This function should be run after get_enriched_go.

estimate_go_overrep(obj, pwf, gene2cat)

Arguments

obj

data.frame containing goseq results as generated by get_enriched_go or goseq.

pwf

data.frame as used in get_enriched_go or goseq.

gene2cat

data.frame as used in get_enriched_go or goseq.

Value

Returns obj with an extra column added called adj_overrep. This column is calculated for each GO term by:

numDEInCat / numInCat / (avgTermWeight / avgNonTermWeight) / (totalDEFeatures / totalFeatures)

where:

  • numDEInCat is the number of differentially expressed genes (aka. proteins) assigned to that GO term.

  • numInCat is the total number of genes (aka. proteins) annotated to that GO term.

  • avgTermWeight is the average pwf$pwf value for all the differentially expressed genes that were assigned to that GO term.

  • avgNonTermWeight is the average pwf$pwf for all the other genes supplied in pwf.

  • totalDEFeatures is the total number of differentially expressed genes indicated in pwf.

  • totalFeatures is the total number of genes indicated in pwf, i.e. the number of rows.