Quantitative structure–activity relationship (QSAR) study of interleukin-1 receptor associated kinase 4 (IRAK-4) inhibitor activity by the genetic algorithm and multiple linear regression (GA-MLR) method

Figures & data

Table 1.  Chemical structures and the corresponding observed and predicted pIC₅₀ values by the MLR method.

No.	General structure	R1	R2	Exp.	Pred.
1		2-Cl	-	7.15	6.50
2		H	-	5.55	5.71
3^a		2-Me	-	5.14	5.36
4		2-OMe	-	5.14	5.28
5		2-OCHF2	-	6.10	5.77
6		3-Me	-	6.00	6.07
7		3-OMe	-	5.32	4.94
8		4-OMe	-	5.05	5.41
9^a		4-Oxazole	-	5.43	5.38
10		4-Piperidine	-	5.19	5.33
11^a		4-(N-Methyl) piperazine	-	5.62	5.76
12		4-Morpholine	-	6.15	6.18
13		2,4-Di-OMe	-	5.55	5.71
14^a		2-OMe,4-Morpholine	-	5.92	5.67
15		2-OMe,4-(N-Methyl) piperazine	-	6.80	7.08
16		-	-	6.38	6.20
17		-	-	7.70	7.49
18			-	5.47	5.76
19			-	5.80	5.66
20			-	6.18	5.66
21			-	6.15	5.94
22			-	7.15	6.50
23			-	6.34	6.70
24			-	5.71	5.61
25^a		H	H	7.15	6.50
26		H	OMe	7.82	6.99
27^a		OMe	H	6.68	6.93
28		OMe	OMe	6.74	7.28
29^a		H	Cl	8.52	8.28
30		Cl	H	6.95	7.04
31		Cl	Cl	6.78	7.16
32		H	Me	8.00	7.33
33		Me	H	5.96	7.16
34			-	5.94	5.82
35^a			-	6.36	6.55
36			-	5.95	6.62
37			-	6.30	6.43
38			-	5.09	5.14
39			-	5.89	5.87
40^a			-	6.04	5.43
41			-	6.65	7.05
42			-	5.52	6.12
43			-	6.50	6.78
44			-	7.41	7.27
45^a			-	6.26	6.93
46^a			-	7.72	6.62
47^a			-	6.90	7.66
48			-	5.73	5.42
49			-	6.27	6.09
50			-	6.31	6.25
51		Cl		8.22	8.75
52		Cl		9.00	8.50
53		Cl		8.05	8.60
54		OMe		8.15	8.41
55		CN		7.02	7.66
56		Cl	-	9.00	8.54
57		OMe	-	8.22	8.07
58		CN	-	8.40	7.61
59^a			-	8.10	8.16
60			-	8.15	8.04
61			-	8.70	7.88
62			-	7.20	7.16
63			-	8.30	7.81
64			-	6.58	6.85
65			-	7.31	7.07

^aTest set

Figure 1.  The principal component analysis of the training and test sets.

Table 2.  The linear model based on the seven parameters selected by the GA-MLR method.

Descriptor	Chemical meaning	MF^a	VIF^b
Constant	Intercept	-	-
MAOEP	Max atomic orbital electronic population	−0.805	1.303
PW5	Path/walk 5 - Randic shape index	−0.453	2.255
MWC09	Molecular walk count of order 09	−0.158	1.823
Mor12m	3D-MoRSE - signal 12/weighted by atomic masses	0.11	1.403
G2p	2st component symmetry directional WHIM index/weighted by atomic polarisabilities	1.768	1.438
R4e	R autocorrelation of lag 4/weighted by atomic Sanderson electronegativities	0.609	1.701
nNHR	Number of secondary amines (aliphatic)	0.071	1.709

^aMean effect

^bVariation inflation factors

Figure 2.  The predicted versus the experimental pIC₅₀ by GA-MLR.

Figure 3.  The residual versus the experimental pIC₅₀ by GA-MLR.

Table 3.  The R²_train and Q²_LOO values after several Y-randomisation tests.

Iteration	R^2train	Q^2LOO
1	0.005	0.173
2	0.053	0.077
3	0.024	0.220
4	0.006	0.178
5	0.064	0.272
6	0.005	0.136
7	0.003	0.141
8	0.020	0.100
9	0.045	0.096
10	0.006	0.110

Figure 4.  The William plot of the GA-MLR model.