Table 2.

Characteristics of the SLR-LS models used in the optimization study.

Set	SLR model	R 2	s	F	n
1a	log⁡(1/IGC50) = +0.677 · log⁡P − 1.38	0.90	0.22	287	35
1b	log⁡(1/IGC50) = +0.647 · log⁡P − 1.05	0.84	0.30	666	126
1c	log⁡(1/IGC50) = −0.443 · log⁡P + 0.509	0.53	0.57	276	250
2	log⁡P = −0.004 · ISDRTHg∗ + 2.09	0.53	0.43	25	24
3	BP = +188.40 · lbMdsHg∗ − 507.95	0.99	3.81	8050	73
4a	log⁡P = +0.99998 · SD + 5.232	0.71	0.32	92	40
4b	ln⁡(LD50) = +0.0018 · SD − 61.168	0.41	0.98	19	30
5	log⁡RBA = +0.026 · TIC1 − 4.145	0.36	1.44	72	132
6	pIC50 = +0.255 · DCW − 1.216	0.71	0.57	191	80
7	log⁡K I = −0.578 · N-rings + 2.646	0.49	0.37	43	47
8	log⁡(1/MC) = −4.129 · TEuIFFDL∗ + 5.789	0.65	0.38	64	37
9	ACE = 47.5480 · IHMdpMg∗ − 0.1687	0.74	0.33	128	47
10	ln⁡(RT) = 0.348 · log⁡P + 1.711	0.56	0.50	75	60

SLR = simple linear regression.

log⁡(1/IGC₅₀) = concentrations (expressed as mM) producing a 50% growth inhibition on T. pyriformis.

^∗MDF descriptors [33, 39, 40, 42].

SD = global correlation descriptor [35]; TIC1 = total information content index (neighborhood symmetry of 1-order).

DCW = flexible (activity dependent) descriptor.

std_dim3 = the square root of the third largest eigenvalue of the covariance matrix of the atomic coordinates [43].

R ² = determination coefficient; s = standard error of the estimate.

F = Fisher's statistic of the regression model; n = sample size.