Benchmarking Structural Inference Methods.

Benchmarking on Raw Trajectories

The results shown in the tables are the average AUROC values of 10 runs on the trajectories generated by a certain type of graphs. The column headers represent the number of nodes in the graph, e.g., “n15” denotes the graph with 15 nodes.

a) AUROC values (in %) of investigated structural inference methods on BN trajectories.

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	96.12 ± 0.40	98.08 ± 0.22	98.83 ± 0.09	99.43 ± 0.01
TIGRESS	-	-	-	-	93.14 ± 0.67	96.44 ± 0.76	97.72 ± 0.26	98.72 ± 0.04
ARACNe	-	-	-	-	94.10 ± 0.66	96.45 ± 0.31	97.78 ± 0.19	98.83 ± 0.03
CLR	-	-	-	-	95.39 ± 0.48	96.72 ± 0.56	97.73 ± 0.19	98.84 ± 0.03
PIDC	-	-	-	-	88.45 ± 0.61	93.16 ± 0.69	94.28 ± 0.26	96.17 ± 0.12
Scribe	-	-	-	-	48.71 ± 1.37	62.41 ± 1.64	68.79 ± 2.53	69.36 ± 1.50
dynGENIE3	-	-	-	-	90.70 ± 2.97	99.87 ± 0.01	99.89 ± 0.00	99.97 ± 0.00
XGBGRN	-	-	-	-	100.00 ± 0.00	100.00 ± 0.00	100.00 ± 0.00	100.00 ± 0.00
NRI	99.75 ± 0.00	99.57 ± 0.00	99.12 ± 0.01	97.54 ± 0.02	99.79 ± 0.00	98.73 ± 0.00	76.08 ± 0.01	75.26 ± 0.01
ACD	99.75 ± 0.00	99.60 ± 0.00	98.96 ± 0.01	99.57 ± 0.01	99.87 ± 0.00	98.95 ± 0.00	80.96 ± 0.01	79.88 ± 0.01
MPM	99.98 ± 0.00	99.95 ± 0.00	99.97 ± 0.01	98.69 ± 0.01	99.95 ± 0.00	99.56 ± 0.00	98.60 ± 0.01	79.92 ± 0.01
iSIDG	99.97 ± 0.00	99.94 ± 0.00	99.95 ± 0.01	98.92 ± 0.01	99.91 ± 0.00	99.62 ± 0.00	98.59 ± 0.01	76.41 ± 0.01
RCSI	99.81 ± 0.01	99.46 ± 0.01	99.50 ± 0.01	98.04 ± 0.01	99.72 ± 0.01	99.43 ± 0.00	98.60 ± 0.01	80.01 ± 0.01

b) AUROC values (in %) of investigated structural inference methods on CRNA trajectories

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	91.37 ± 1.10	90.35 ± 0.39	90.26 ± 0.54	89.16 ± 0.55
TIGRESS	-	-	-	-	84.40 ± 2.84	74.88 ± 0.64	69.41 ± 0.64	60.10 ± 0.46
ARACNe	-	-	-	-	78.11 ± 1.50	77.93 ± 1.00	77.55 ± 0.80	75.74 ± 0.89
CLR	-	-	-	-	86.01 ± 1.98	86.59 ± 1.06	84.24 ± 0.76	81.14 ± 1.24
PIDC	-	-	-	-	85.70 ± 3.35	75.38 ± 0.42	70.81 ± 1.99	82.74 ± 0.88
Scribe	-	-	-	-	55.19 ± 3.80	52.19 ± 0.22	50.78 ± 0.25	50.94 ± 0.74
dynGENIE3	-	-	-	-	56.92 ± 6.83	50.32 ± 1.36	50.12 ± 0.84	50.35 ± 0.60
XGBGRN	-	-	-	-	99.60 ± 0.30	99.58 ± 0.13	97.40 ± 0.52	51.48 ± 0.22
NRI	83.91 ± 0.03	72.81 ± 0.05	70.73 ± 0.02	65.32 ± 0.02	49.47 ± 0.02	49.03 ± 0.03	50.06 ± 0.02	50.65 ± 0.02
ACD	85.90 ± 0.04	75.41 ± 0.01	69.97 ± 0.01	64.51 ± 0.02	48.26 ± 0.02	48.40 ± 0.03	51.42 ± 0.01	50.21 ± 0.02
MPM	85.75 ± 0.03	73.71 ± 0.01	68.25 ± 0.02	64.87 ± 0.02	49.72 ± 0.01	51.16 ± 0.04	50.06 ± 0.01	50.56 ± 0.02
iSIDG	87.01 ± 0.02	78.21 ± 0.05	70.72 ± 0.01	62.31 ± 0.02	51.04 ± 0.01	50.24 ± 0.04	51.26 ± 0.01	50.87 ± 0.02
RCSI	87.51 ± 0.02	78.11 ± 0.05	69.82 ± 0.02	64.80 ± 0.03	51.15 ± 0.02	50.81 ± 0.04	51.10 ± 0.02	50.00 ± 0.02

c) AUROC values (in %) of investigated structural inference methods on FW trajectories.

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	78.21 ± 1.57	73.63 ± 1.75	72.76 ± 1.04	71.72 ± 0.15
TIGRESS	-	-	-	-	64.15 ± 1.55	58.00 ± 0.61	57.92 ± 0.84	53.97 ± 0.44
ARACNe	-	-	-	-	66.07 ± 4.26	65.40 ± 3.82	68.39 ± 0.24	53.18 ± 2.03
CLR	-	-	-	-	79.69 ± 3.33	74.20 ± 1.57	73.94 ± 1.01	44.50 ± 2.24
PIDC	-	-	-	-	78.82 ± 3.75	50.00 ± 0.00	50.00 ± 0.00	64.72 ± 1.39
Scribe	-	-	-	-	52.96 ± 2.66	54.25 ± 1.16	51.02 ± 1.59	51.73 ± 0.92
dynGENIE3	-	-	-	-	47.98 ± 2.67	49.89 ± 1.29	49.40 ± 0.58	51.26 ± 1.07
XGBGRN	-	-	-	-	84.84 ± 1.90	73.00 ± 4.00	52.36 ± 0.35	49.11 ± 0.77
NRI	81.80 ± 0.01	76.75 ± 0.02	74.15 ± 0.01	71.57 ± 0.01	49.30 ± 0.03	48.50 ± 0.03	50.75 ± 0.02	47.56 ± 0.03
ACD	81.89 ± 0.01	76.38 ± 0.02	73.50 ± 0.01	71.12 ± 0.01	50.74 ± 0.06	50.19 ± 0.01	50.49 ± 0.03	49.82 ± 0.01
MPM	81.87 ± 0.02	75.97 ± 0.01	73.59 ± 0.01	71.52 ± 0.01	53.01 ± 0.08	50.66 ± 0.01	51.22 ± 0.03	53.01 ± 0.03
iSIDG	81.95 ± 0.01	76.75 ± 0.01	74.38 ± 0.02	72.21 ± 0.02	53.36 ± 0.03	50.78 ± 0.03	50.46 ± 0.03	51.07 ± 0.01
RCSI	81.80 ± 0.01	75.62 ± 0.02	74.51 ± 0.02	72.40 ± 0.02	53.70 ± 0.05	50.28 ± 0.03	51.61 ± 0.03	53.82 ± 0.02

d) AUROC values (in %) of investigated structural inference methods on GCN trajectories.

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	96.72 ± 1.64	98.48 ± 0.35	98.55 ± 0.22	98.20 ± 0.42
TIGRESS	-	-	-	-	91.72 ± 4.28	87.90 ± 1.44	80.44 ± 1.78	78.12 ± 0.15
ARACNe	-	-	-	-	95.24 ± 0.00	91.15 ± 2.18	92.75 ± 1.94	94.04 ± 0.71
CLR	-	-	-	-	94.57 ± 2.14	97.48 ± 0.54	97.25 ± 0.76	96.40 ± 0.21
PIDC	-	-	-	-	92.75 ± 3.92	91.98 ± 0.90	92.01 ± 1.23	94.17 ± 1.25
Scribe	-	-	-	-	50.47 ± 2.55	49.31 ± 1.72	48.17 ± 2.80	49.51 ± 0.77
dynGENIE3	-	-	-	-	46.70 ± 5.05	47.86 ± 4.04	50.46 ± 1.93	49.58 ± 1.37
XGBGRN	-	-	-	-	93.28 ± 2.47	96.71 ± 0.51	96.74 ± 0.62	94.95 ± 0.33
NRI	97.42 ± 0.00	93.38 ± 0.01	89.54 ± 0.02	83.78 ± 0.01	43.46 ± 0.02	52.74 ± 0.06	50.98 ± 0.02	50.34 ± 0.02
ACD	97.95 ± 0.01	92.62 ± 0.01	89.96 ± 0.03	90.73 ± 0.02	42.23 ± 0.03	46.12 ± 0.04	47.66 ± 0.03	49.87 ± 0.04
MPM	98.82 ± 0.01	92.68 ± 0.02	85.81 ± 0.03	84.98 ± 0.02	52.59 ± 0.03	66.65 ± 0.07	63.01 ± 0.07	53.07 ± 0.06
iSIDG	98.93 ± 0.01	93.16 ± 0.01	89.53 ± 0.01	87.60 ± 0.01	56.41 ± 0.06	52.07 ± 0.03	52.96 ± 0.03	50.78 ± 0.03
RCSI	97.66 ± 0.01	93.92 ± 0.02	88.69 ± 0.02	88.01 ± 0.02	53.31 ± 0.05	59.23 ± 0.03	54.30 ± 0.02	50.44 ± 0.02

e) AUROC values (in %) of investigated structural inference methods on GRN trajectories.

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	86.12 ± 0.98	88.72 ± 1.33	89.83 ± 0.89	89.61 ± 0.93
TIGRESS	-	-	-	-	79.09 ± 1.07	85.16 ± 2.26	85.85 ± 1.96	87.41 ± 2.73
ARACNe	-	-	-	-	70.46 ± 3.52	70.05 ± 2.10	70.73 ± 1.90	69.48 ± 2.10
CLR	-	-	-	-	78.25 ± 0.49	76.48 ± 1.91	75.67 ± 1.29	73.09 ± 2.10
PIDC	-	-	-	-	57.49 ± 3.59	63.51 ± 2.69	65.95 ± 1.41	63.85 ± 2.00
Scribe	-	-	-	-	44.89 ± 7.52	47.79 ± 3.50	45.50 ± 3.03	46.15 ± 2.41
dynGENIE3	-	-	-	-	64.23 ± 4.75	59.69 ± 6.09	54.38 ± 3.18	58.53 ± 3.94
XGBGRN	-	-	-	-	80.08 ± 3.81	83.77 ± 0.49	84.51 ± 0.43	83.47 ± 1.31
NRI	91.65 ± 0.01	90.45 ± 0.01	90.35 ± 0.02	88.14 ± 0.02	78.08 ± 0.03	57.01 ± 0.05	55.71 ± 0.05	58.33 ± 0.04
ACD	91.10 ± 0.00	88.21 ± 0.01	86.78 ± 0.01	90.07 ± 0.03	80.18 ± 0.04	69.78 ± 0.07	62.65 ± 0.02	53.99 ± 0.03
MPM	94.02 ± 0.01	93.25 ± 0.02	84.60 ± 0.02	85.30 ± 0.02	70.46 ± 0.04	57.36 ± 0.03	72.25 ± 0.05	66.74 ± 0.03
iSIDG	92.91 ± 0.01	90.06 ± 0.01	90.15 ± 0.01	87.94 ± 0.04	71.11 ± 0.04	56.25 ± 0.02	57.15 ± 0.02	62.13 ± 0.02
RCSI	93.88 ± 0.02	93.01 ± 0.02	90.35 ± 0.01	89.90 ± 0.03	77.45 ± 0.03	65.77 ± 0.03	59.93 ± 0.02	60.15 ± 0.03

f) AUROC values (in %) of investigated structural inference methods on IN trajectories.

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	94.14 ± 0.54	96.13 ± 1.65	97.64 ± 0.11	97.61 ± 0.01
TIGRESS	-	-	-	-	94.39 ± 1.34	91.79 ± 5.82	86.31 ± 1.42	78.25 ± 0.52
ARACNe	-	-	-	-	85.82 ± 3.90	87.77 ± 5.36	83.05 ± 2.84	86.14 ± 0.54
CLR	-	-	-	-	87.17 ± 0.26	92.45 ± 2.50	89.58 ± 3.93	92.82 ± 1.03
PIDC	-	-	-	-	81.90 ± 1.92	85.16 ± 1.59	84.84 ± 2.89	89.35 ± 0.48
Scribe	-	-	-	-	54.29 ± 4.17	50.81 ± 1.34	50.68 ± 3.52	50.76 ± 0.53
dynGENIE3	-	-	-	-	58.18 ± 4.97	70.18 ± 15.42	68.08 ± 8.25	50.22 ± 1.78
XGBGRN	-	-	-	-	99.00 ± 0.85	99.69 ± 0.07	99.90 ± 0.04	99.91 ± 0.05
NRI	93.09 ± 0.01	90.54 ± 0.05	88.10 ± 0.03	82.51 ± 0.03	60.47 ± 0.04	61.78 ± 0.06	56.45 ± 0.04	53.96 ± 0.04
ACD	93.33 ± 0.02	89.12 ± 0.05	87.69 ± 0.04	81.37 ± 0.02	68.39 ± 0.06	55.11 ± 0.08	53.88 ± 0.02	53.04 ± 0.05
MPM	95.61 ± 0.02	89.59 ± 0.05	86.47 ± 0.03	83.45 ± 0.03	63.83 ± 0.03	64.70 ± 0.09	54.18 ± 0.03	54.37 ± 0.04
iSIDG	95.37 ± 0.02	90.72 ± 0.05	87.79 ± 0.02	84.00 ± 0.02	62.18 ± 0.03	61.91 ± 0.01	56.50 ± 0.02	53.85 ± 0.02
RCSI	96.70 ± 0.01	93.46 ± 0.02	88.02 ± 0.02	83.96 ± 0.01	62.08 ± 0.04	60.05 ± 0.02	55.65 ± 0.03	52.84 ± 0.02

g) AUROC values (in %) of investigated structural inference methods on LN trajectories

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	99.49 ± 0.56	95.04 ± 5.20	86.75 ± 1.66	79.32 ± 4.32
TIGRESS	-	-	-	-	84.15 ± 1.16	87.38 ± 3.32	92.22 ± 0.42	93.97 ± 1.96
ARACNe	-	-	-	-	92.33 ± 4.84	80.36 ± 5.67	71.17 ± 0.48	62.82 ± 8.36
CLR	-	-	-	-	97.35 ± 3.17	96.56 ± 4.87	91.04 ± 2.35	95.04 ± 0.53
PIDC	-	-	-	-	97.53 ± 1.01	82.03 ± 7.28	88.58 ± 1.69	94.18 ± 2.28
Scribe	-	-	-	-	54.22 ± 3.98	56.16 ± 3.88	52.12 ± 2.49	52.55 ± 1.62
dynGENIE3	-	-	-	-	51.32 ± 5.21	50.12 ± 2.42	50.49 ± 1.22	67.32 ± 14.23
XGBGRN	-	-	-	-	97.21 ± 1.13	96.95 ± 2.10	96.90 ± 0.83	97.99 ± 0.93
NRI	97.01 ± 0.02	94.94 ± 0.00	87.10 ± 0.01	82.80 ± 0.01	56.00 ± 0.04	53.94 ± 0.02	54.36 ± 0.02	51.75 ± 0.03
ACD	96.99 ± 0.02	95.79 ± 0.01	87.58 ± 0.02	83.92 ± 0.02	61.94 ± 0.03	61.56 ± 0.04	53.36 ± 0.02	50.19 ± 0.02
MPM	97.92 ± 0.01	95.53 ± 0.02	86.92 ± 0.01	84.22 ± 0.03	52.18 ± 0.02	62.08 ± 0.05	53.44 ± 0.01	50.42 ± 0.03
iSIDG	97.38 ± 0.02	94.70 ± 0.02	87.44 ± 0.02	83.15 ± 0.02	59.19 ± 0.05	56.18 ± 0.03	55.73 ± 0.03	52.30 ± 0.02
RCSI	97.30 ± 0.02	94.42 ± 0.02	88.02 ± 0.02	84.26 ± 0.02	60.28 ± 0.03	60.15 ± 0.02	57.56 ± 0.03	53.48 ± 0.02

h) AUROC values (in %) of investigated structural inference methods on MMO trajectories

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	96.42 ± 0.02	98.28 ± 0.00	98.98 ± 0.00	99.49 ± 0.00
TIGRESS	-	-	-	-	99.88 ± 0.01	99.98 ± 0.00	100.00 ± 0.00	100.00 ± 0.00
ARACNe	-	-	-	-	89.76 ± 0.16	96.60 ± 1.51	97.09 ± 1.07	98.11 ± 0.79
CLR	-	-	-	-	96.43 ± 0.00	98.28 ± 0.00	98.98 ± 0.00	98.81 ± 0.37
PIDC	-	-	-	-	44.74 ± 4.70	70.03 ± 7.65	77.24 ± 1.02	75.01 ± 0.29
Scribe	-	-	-	-	69.85 ± 12.21	38.03 ± 25.86	20.70 ± 10.19	23.88 ± 15.76
dynGENIE3	-	-	-	-	16.90 ± 2.38	23.49 ± 5.12	23.31 ± 4.03	45.89 ± 20.23
XGBGRN	-	-	-	-	59.77 ± 2.14	81.64 ± 6.68	72.13 ± 11.09	63.83 ± 6.71
NRI	99.62 ± 0.00	84.96 ± 0.02	77.66 ± 0.01	78.04 ± 0.02	68.34 ± 0.03	66.21 ± 0.06	57.84 ± 0.03	56.10 ± 0.01
ACD	99.68 ± 0.00	93.89 ± 0.01	85.53 ± 0.02	85.46 ± 0.01	71.88 ± 0.03	59.46 ± 0.06	64.14 ± 0.03	58.05 ± 0.02
MPM	99.83 ± 0.00	88.32 ± 0.01	87.02 ± 0.03	86.75 ± 0.02	79.34 ± 0.04	65.48 ± 0.07	54.78 ± 0.04	57.06 ± 0.02
iSIDG	99.84 ± 0.00	89.77 ± 0.01	87.47 ± 0.02	85.47 ± 0.01	74.58 ± 0.03	64.71 ± 0.06	56.07 ± 0.04	58.80 ± 0.01
RCSI	99.70 ± 0.01	92.73 ± 0.02	88.05 ± 0.02	85.49 ± 0.02	73.61 ± 0.04	66.08 ± 0.02	57.90 ± 0.03	58.74 ± 0.02

i) AUROC values (in %) of investigated structural inference methods on RNLO trajectories

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	96.36 ± 0.10	98.28 ± 0.00	98.95 ± 0.04	99.25 ± 0.38
TIGRESS	-	-	-	-	99.82 ± 0.06	99.98 ± 0.00	99.99 ± 0.00	99.99 ± 0.01
ARACNe	-	-	-	-	93.47 ± 2.99	95.67 ± 1.61	97.02 ± 0.86	98.03 ± 0.43
CLR	-	-	-	-	96.35 ± 0.12	98.28 ± 0.00	98.72 ± 0.31	98.62 ± 0.29
PIDC	-	-	-	-	56.18 ± 6.51	72.67 ± 10.76	74.36 ± 6.83	71.95 ± 2.31
Scribe	-	-	-	-	38.49 ± 1.57	47.15 ± 18.16	46.52 ± 26.84	20.23 ± 13.56
dynGENIE3	-	-	-	-	15.96 ± 2.97	21.37 ± 8.84	27.57 ± 7.69	56.44 ± 21.63
XGBGRN	-	-	-	-	83.55 ± 8.24	81.05 ± 5.42	81.82 ± 5.07	67.30 ± 12.31
NRI	95.54 ± 0.02	72.53 ± 0.08	72.72 ± 0.03	75.07 ± 0.02	69.43 ± 0.04	67.70 ± 0.08	60.55 ± 0.03	62.42 ± 0.02
ACD	96.20 ± 0.02	93.44 ± 0.03	75.83 ± 0.02	79.14 ± 0.02	57.32 ± 0.05	53.75 ± 0.01	61.68 ± 0.05	65.45 ± 0.03
MPM	97.40 ± 0.01	83.70 ± 0.06	78.50 ± 0.02	79.36 ± 0.02	72.62 ± 0.03	62.34 ± 0.01	56.90 ± 0.05	60.05 ± 0.02
iSIDG	97.45 ± 0.01	81.60 ± 0.05	78.51 ± 0.03	79.08 ± 0.03	64.79 ± 0.05	57.10 ± 0.02	64.50 ± 0.05	66.01 ± 0.02
RCSI	97.30 ± 0.01	83.05 ± 0.03	80.43 ± 0.02	79.04 ± 0.02	69.92 ± 0.04	59.42 ± 0.03	60.99 ± 0.04	60.24 ± 0.02

j) AUROC values (in %) of investigated structural inference methods on SN trajectories

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	93.77 ± 0.59	94.17 ± 0.28	94.74 ± 0.44	94.37 ± 0.05
TIGRESS	-	-	-	-	90.20 ± 1.52	82.82 ± 0.30	78.22 ± 1.92	67.98 ± 0.57
ARACNe	-	-	-	-	80.80 ± 3.58	78.78 ± 3.00	80.42 ± 1.00	81.49 ± 0.32
CLR	-	-	-	-	85.08 ± 0.54	87.70 ± 1.11	89.81 ± 0.74	88.24 ± 0.60
PIDC	-	-	-	-	83.96 ± 2.44	84.29 ± 1.00	84.66 ± 0.70	91.76 ± 0.25
Scribe	-	-	-	-	56.52 ± 2.94	51.30 ± 0.50	50.38 ± 0.50	50.74 ± 1.01
dynGENIE3	-	-	-	-	62.48 ± 5.44	55.74 ± 3.23	50.00 ± 1.70	50.20 ± 0.77
XGBGRN	-	-	-	-	99.83 ± 0.21	99.88 ± 0.07	99.74 ± 0.12	98.81 ± 0.12
NRI	93.26 ± 0.01	79.96 ± 0.02	80.40 ± 0.02	71.84 ± 0.01	58.41 ± 0.04	51.43 ± 0.01	49.57 ± 0.03	50.16 ± 0.03
ACD	93.47 ± 0.01	81.17 ± 0.01	79.63 ± 0.02	68.76 ± 0.02	65.24 ± 0.05	52.96 ± 0.03	49.28 ± 0.02	50.76 ± 0.01
MPM	92.68 ± 0.00	79.32 ± 0.01	75.90 ± 0.01	69.36 ± 0.03	67.42 ± 0.02	50.87 ± 0.01	53.12 ± 0.03	50.08 ± 0.02
iSIDG	93.51 ± 0.00	81.38 ± 0.01	80.80 ± 0.02	69.25 ± 0.01	66.14 ± 0.04	53.79 ± 0.03	54.83 ± 0.01	51.72 ± 0.02
RCSI	94.13 ± 0.02	82.66 ± 0.01	81.21 ± 0.01	73.42 ± 0.02	67.58 ± 0.03	55.84 ± 0.02	55.04 ± 0.02	53.24 ± 0.03

k) AUROC values (in %) of investigated structural inference methods on VN trajectories

Method	Springs				NetSims
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	-	-	-	-	96.68 ± 0.01	98.33 ± 0.01	99.00 ± 0.00	99.50 ± 0.00
TIGRESS	-	-	-	-	99.28 ± 0.18	99.41 ± 0.15	99.62 ± 0.09	99.84 ± 0.02
ARACNe	-	-	-	-	96.66 ± 0.03	97.85 ± 0.09	98.54 ± 0.01	99.08 ± 0.00
CLR	-	-	-	-	96.68 ± 0.00	98.34 ± 0.00	99.00 ± 0.00	99.50 ± 0.00
PIDC	-	-	-	-	76.51 ± 2.67	85.70 ± 3.99	91.80 ± 0.43	95.01 ± 0.70
Scribe	-	-	-	-	51.56 ± 5.64	52.71 ± 4.98	57.68 ± 2.56	59.50 ± 0.83
dynGENIE3	-	-	-	-	92.81 ± 2.83	97.33 ± 1.01	97.87 ± 0.66	97.30 ± 1.26
XGBGRN	-	-	-	-	97.99 ± 0.49	98.54 ± 0.38	99.21 ± 0.12	99.59 ± 0.02
NRI	94.58 ± 0.01	95.12 ± 0.01	94.65 ± 0.02	89.17 ± 0.02	90.31 ± 0.01	74.64 ± 0.04	69.78 ± 0.03	68.80 ± 0.02
ACD	94.34 ± 0.01	93.73 ± 0.01	87.54 ± 0.03	90.49 ± 0.03	80.32 ± 0.02	65.36 ± 0.06	69.01 ± 0.03	68.72 ± 0.03
MPM	96.56 ± 0.01	89.71 ± 0.04	85.07 ± 0.02	84.56 ± 0.03	91.18 ± 0.01	83.37 ± 0.03	72.66 ± 0.04	70.34 ± 0.03
iSIDG	96.59 ± 0.02	95.66 ± 0.01	95.72 ± 0.02	85.07 ± 0.02	91.20 ± 0.02	78.08 ± 0.06	73.68 ± 0.02	68.81 ± 0.02
RCSI	97.03 ± 0.01	95.31 ± 0.01	94.48 ± 0.02	90.72 ± 0.03	91.53 ± 0.02	82.27 ± 0.04	74.08 ± 0.02	70.29 ± 0.03

Benchmarking on EMT Dataset

The results shown in the table are the average AUROC values of 10 runs on the EMT trajectories.

AUROC values (in %) of investigated structural inference methods on EMT dataset.

Method	AUROC
ppcor	55.31 ± 0.00
TIGRESS	56.32 ± 0.28
ARACNe	57.22 ± 0.00
CLR	51.41 ± 0.00
PIDC	54.53 ± 0.00
Scribe	54.82 ± 0.00
dynGENIE3	44.42 ± 0.05
XGBGRN	55.63 ± 0.74
NRI	52.09 ± 0.06
ACD	51.14 ± 0.03
MPM	52.43 ± 0.07
iSIDG	52.58 ± 0.06
RCSI	53.02 ± 0.07

Benchmarking on Trajectories with Noise

The results shown in the tables are the average AUROC values of 10 runs on the trajectories generated by a BN trajectories with a certain level of noise. The column headers represent the level of noise, e.g., “N1” denotes the trajectories have one level of added Gaussian noise.

a) AUROC values (in %) of investigated structural inference methods on BN trajectories with 1 (N1) and 2 (N2) levels of Gaussian noise.

Method	N1				N2
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	92.66 ± 0.80	97.16 ± 0.59	98.48 ± 0.19	99.30 ± 0.02	91.25 ± 0.75	96.68 ± 0.64	98.28 ± 0.22	99.21 ± 0.03
TIGRESS	93.08 ± 0.76	96.42 ± 0.67	97.59 ± 0.23	98.65 ± 0.05	93.12 ± 0.80	96.43 ± 0.62	97.55 ± 0.24	98.59 ± 0.05
ARACNe	84.73 ± 1.20	91.90 ± 1.00	95.84 ± 0.33	98.11 ± 0.11	84.39 ± 1.04	92.37 ± 0.98	95.73 ± 0.34	97.76 ± 0.13
CLR	91.46 ± 0.45	96.48 ± 0.64	97.97 ± 0.24	98.97 ± 0.03	90.88 ± 0.73	96.55 ± 0.67	98.12 ± 0.20	99.04 ± 0.03
PIDC	87.87 ± 0.64	94.54 ± 0.41	95.84 ± 0.10	96.77 ± 0.08	88.58 ± 0.66	95.02 ± 0.75	96.78 ± 0.19	97.56 ± 0.06
Scribe	47.75 ± 6.78	63.04 ± 2.33	73.37 ± 1.11	70.95 ± 1.87	46.19 ± 5.58	63.42 ± 4.19	72.37 ± 1.98	71.36 ± 1.12
dynGENIE3	83.60 ± 3.35	90.28 ± 1.63	92.28 ± 2.10	98.00 ± 0.45	76.46 ± 0.64	88.32 ± 3.03	90.96 ± 1.39	97.93 ± 0.04
XGBGRN	93.72 ± 1.08	98.35 ± 0.21	98.63 ± 0.18	99.40 ± 0.01	86.78 ± 2.19	96.92 ± 1.00	97.94 ± 0.28	99.07 ± 0.05
NRI	72.98 ± 0.01	73.85 ± 0.02	74.12 ± 0.02	74.70 ± 0.02	56.76 ± 0.02	59.64 ± 0.03	62.52 ± 0.03	63.52 ± 0.02
ACD	65.62 ± 0.02	63.47 ± 0.01	66.69 ± 0.02	61.56 ± 0.03	62.08 ± 0.02	58.14 ± 0.03	61.73 ± 0.02	59.04 ± 0.02
MPM	70.23 ± 0.02	74.37 ± 0.02	75.72 ± 0.03	75.60 ± 0.03	62.83 ± 0.02	65.22 ± 0.02	66.52 ± 0.02	66.88 ± 0.03
iSIDG	74.33 ± 0.03	76.06 ± 0.02	76.29 ± 0.01	76.54 ± 0.03	63.40 ± 0.04	66.44 ± 0.03	67.52 ± 0.03	68.75 ± 0.02
RCSI	73.09 ± 0.03	74.50 ± 0.03	76.83 ± 0.02	76.01 ± 0.02	63.90 ± 0.02	64.72 ± 0.02	65.31 ± 0.03	66.62 ± 0.02

b) AUROC values (in %) of investigated structural inference methods on BN trajectories with 3 (N3) and 4 (N4) levels of Gaussian noise.

Method	N3				N4
Method		n15	n30	n50	n100	n15	n30	n50	n100
ppcor	90.87 ± 0.66	96.36 ± 0.62	98.16 ± 0.19	99.15 ± 0.04	90.81 ± 0.67	96.10 ± 0.65	98.09 ± 0.19	99.09 ± 0.04
TIGRESS	93.11 ± 0.65	96.45 ± 0.62	97.59 ± 0.21	98.56 ± 0.05	93.00 ± 0.38	96.44 ± 0.60	97.64 ± 0.22	98.57 ± 0.05
ARACNe	88.04 ± 1.01	93.42 ± 0.85	96.02 ± 0.34	97.80 ± 0.11	89.51 ± 0.73	93.89 ± 0.79	96.22 ± 0.35	97.85 ± 0.12
CLR	91.22 ± 0.82	96.57 ± 0.70	98.20 ± 0.20	99.07 ± 0.03	91.40 ± 0.86	96.63 ± 0.71	98.26 ± 0.20	99.09 ± 0.03
PIDC	90.24 ± 0.56	95.17 ± 0.75	96.93 ± 0.23	97.98 ± 0.04	91.53 ± 1.11	95.17 ± 0.84	97.03 ± 0.28	98.12 ± 0.04
Scribe	51.12 ± 2.82	61.51 ± 3.27	71.40 ± 3.26	72.10 ± 0.97	48.14 ± 2.15	60.82 ± 2.68	67.96 ± 2.52	70.71 ± 1.81
dynGENIE3	63.28 ± 2.16	80.56 ± 2.28	87.03 ± 2.73	98.04 ± 0.03	52.46 ± 0.55	73.68 ± 1.60	81.89 ± 4.03	97.77 ± 0.01
XGBGRN	86.90 ± 1.19	96.38 ± 1.00	97.55 ± 0.32	98.88 ± 0.06	85.29 ± 0.62	95.74 ± 1.21	97.37 ± 0.31	98.75 ± 0.07
NRI	50.67 ± 0.02	51.68 ± 0.01	54.40 ± 0.02	58.16 ± 0.02	50.91 ± 0.03	51.11 ± 0.02	51.24 ± 0.02	52.89 ± 0.03
ACD	50.09 ± 0.03	54.38 ± 0.02	56.42 ± 0.01	56.12 ± 0.02	51.89 ± 0.02	54.65 ± 0.02	55.73 ± 0.03	55.02 ± 0.03
MPM	55.29 ± 0.03	56.81 ± 0.03	57.41 ± 0.02	59.23 ± 0.02	55.85 ± 0.03	57.48 ± 0.01	59.76 ± 0.02	59.90 ± 0.02
iSIDG	56.73 ± 0.02	56.79 ± 0.02	57.71 ± 0.01	60.60 ± 0.03	54.59 ± 0.04	57.82 ± 0.03	58.08 ± 0.02	59.70 ± 0.02
RCSI	54.20 ± 0.02	54.72 ± 0.02	56.44 ± 0.02	59.43 ± 0.03	52.47 ± 0.03	53.02 ± 0.03	59.50 ± 0.02	58.34 ± 0.03

c) AUROC values (in %) of investigated structural inference methods on BN trajectories with 5 level of Gaussian noise.

Method	N5
Method		n15	n30	n50	n100
ppcor	91.11 ± 0.69	95.81 ± 0.61	97.97 ± 0.18	99.04 ± 0.05
TIGRESS	92.95 ± 0.42	96.38 ± 0.64	97.66 ± 0.18	98.57 ± 0.05
ARACNe	90.22 ± 0.96	94.15 ± 0.70	96.33 ± 0.35	97.90 ± 0.11
CLR	91.59 ± 0.90	96.65 ± 0.70	98.31 ± 0.20	99.10 ± 0.04
PIDC	91.18 ± 1.61	95.11 ± 0.95	96.90 ± 0.32	98.17 ± 0.03
Scribe	52.20 ± 6.61	58.31 ± 2.98	66.41 ± 2.87	69.35 ± 1.47
dynGENIE3	47.84 ± 1.10	67.07 ± 2.68	74.14 ± 4.26	97.46 ± 0.03
XGBGRN	85.18 ± 0.34	95.41 ± 1.22	97.27 ± 0.28	98.70 ± 0.08
NRI	46.68 ± 0.03	46.70 ± 0.02	49.57 ± 0.03	49.79 ± 0.03
ACD	46.21 ± 0.03	46.34 ± 0.05	44.06 ± 0.02	44.41 ± 0.02
MPM	55.39 ± 0.05	58.87 ± 0.02	59.07 ± 0.03	60.45 ± 0.03
iSIDG	55.59 ± 0.03	58.82 ± 0.03	59.08 ± 0.01	60.70 ± 0.02
RCSI	52.00 ± 0.04	55.31 ± 0.02	57.43 ± 0.02	58.10 ± 0.03

Benchmarking over Data-Efficiency

Average AUROC results with trajectories of different lengths

On this page