Skip to main content

Table 2 Higher-order metrics arising from combinations of the basic measures documented in Table 1

From: Beyond differential expression: the quest for causal mutations and effector molecules

Measure Algebra formulae Description Example in skeletal muscle context
Phenotype Impact Factor P I F i = 1 2 E i , A + E i , B d E i = A i d E i Average (normalized) expression of the i-th gene across the two conditions multiplied by its differential expression. In other words, PIF weights the differential expression of a given gene by its overall abundance. MYL2 very strongly.
Regulatory Impact Factor, Option 1 R I F 1 i = 1 n dE j = 1 j = n dE P I F j d C i , j 2 For the i-th regulator and across all the j differentially expressed genes (j = 1, …, n dE ) RIF1 looks at the average PIF of the i-th regulator weighted by the squared differential co-expression between the i-th regulator and the j-th differentially expressed gene. It addresses the question: Which regulator is consistently highly differentially co-expressed with the abundant differentially expressed gene? MSTN very strongly.
Regulatory Impact Factor, Option 2 R I F 2 i = 1 n dE j = 1 j = n dE E j , A r A i , j 2 E j , B r B i , j 2 For the i-th regulator and across all the j differentially expressed genes (j = 1, …, n dE ) RIF1 looks at the average change in predictive ability of the i-th regulator to predict the abundance of the j-th differentially expressed gene. It addresses the question: Which regulator has the most altered ability to predict the abundance of differentially expressed genes. MSTN very strongly.