Skip to main content

Table 4 Comparison of the qualitative and quantitative behaviors of the network “perturbation” to the expected behavior in the various datasets and studies

From: Quantification of biological network perturbations for mechanistic insight and diagnostics using two-layer causal models

Category

Gene-level

Enrichment

28-day rat inhalation

28-day rat inhalation

RLAK study

RLAK study

CDKi study

Complete match with

 

statistics

statistics

TNF-IL1-NFKB

xenobiotic

TNF-IL1-NFKB

Xenobiotic

cell cycle

biological expectations

Forward FCS

t

mean

×/

×/

/()

./ ×

×/

0

maxmean

×/ ×

()/

×/ ×

./ ×

×/

0

GSEA

×/ ×

×/

×/()

./()

×/

0

fc

mean

()/

()/

()/

./ ×

×/

0

maxmean

/

/

()/()

./ ×

×/

2

GSEA

×/

×/

×/()

./ ×

×/

0

Backward ORA

fdr

RCR

×/ ×

/ ×

×/ ×

./

/ ×

1

Backward FCS

t

mean

×/ ×

×/

×/ ×

./

/()

1

 

maxmean

×/ ×

×/()

×/ ×

./

/ ×

1

t

GSEA

×/ ×

×/

×/ ×

./

/ ×

1

fc

mean

()/()

/()

/()

./

/()

1

maxmean

/

/ ×

/

./

/()

3

GSEA

×/

×/

×/ ×

./

/ ×

1

NPA

()/

na

/

na

na

1

Backward PT

fc

TopoNPA

/

/

/

./

/

5

  1. A indicates that the result corresponds to the biological expectation, a () indicate that up to one exception it matches the expectation, a × indicates that it does not match the expectation, and “.” means that the aspect is unimportant in the case-study. Finally, “na” indicates that the method does not apply. Each use case is assessed for its quantitative (the first symbol) and its qualitative (second symbol) aspects. The comparison scheme is discussed in the material and methods and the table is derived from Figures S1-S5. Overall, TopoNPA behaves as expected and is the only method fully matching the expectations.