Skip to main content

Table 4 Statistics on edges from inferred SysGenSIM networks

From: Reconstruction of large-scale regulatory networks based on perturbation graphs and transitive reduction: improved methods and their evaluation

 

K ≃ 1.5

K ≃ 2

K ≃ 2.5

Inference algorithm

Edges

TPs

FPs

Edges

TPs

FPs

Edges

TPs

FPs

PGnew

14353

6239

8114

20477

7422

13055

27496

8258

19239

PGnew + TRANSWESDu,w,∞

-15.60%

-0.57%

-27.16%

-15.71%

-1.96%

-23.52%

-14.89%

-3.48%

-19.79%

PGnew + TRANSWESDs,w,∞

-15.35%

-0.40%

-26.83%

-15.17%

-1.60%

-22.89%

-14.44%

-3.04%

-19.34%

PGnew + TRANSWESDu,w,2

-37.47%

-16.01%

-53.97%

-42.82%

-25.03%

-52.93%

-46.69%

-33.13%

-52.51%

PGnew + TRANSWESDs,w,2

-37.24%

-15.65%

-53.85%

-42.38%

-24.17%

-52.73%

-46.09%

-31.77%

-52.23%

PGnew + LTRu,u

-32.06%

-9.68%

-49.27%

-35.28%

-15.07%

-46.77%

-37.67%

-19.83%

-45.33%

PGnew + LTRs,u

-31.93%

-9.51%

-49.17%

-35.03%

-14.61%

-46.64%

-37.36%

-19.20%

-45.16%

PGnew + LTRu,w

-29.74%

-5.76%

-48.18%

-30.64%

-7.79%

-43.63%

-28.62%

-8.87%

-37.10%

PGnew + LTRS,W

-29.64%

-5.66%

-48.09%

-30.49%

-7.53%

-43.54%

-28.45%

-8.53%

-37.00%

  1. The number of edges in the perturbation graph and the number of FPs and TPs are shown in the table. The relative reduction of these measures (edges, TPs and FPs) in the graphs obtained after applying the different TR techniques compared to the PG is also displayed. These averaged statistics are computed from the analysis of the graphs obtained after inferring the 30 SysGenSIM networks simulated according to noise configuration 1 (LL; see Table 2). Thresholds used are β = 2.0 and γ = 0.05 for generating PGnew, α = 0.95 for TRANSWESD·,w,∞, α = 1.50 for TRANSWESD·,w,2 and α = 0.15 for LTR·,w. Analogous tables for the other 8 noise configurations (2, …, 9) are shown in Tables T3–T10 in Additional file 1.