Table 2. Classic model fit (E20+E6+FSR+FSHP+AA) including parameter estimates, corresponding standard errors, and z and p values based on the Wald test.

Descriptors included are sequence entropies E20 and E6, the amino acid qualitative predictor (AA) with 20 classes, and FSHP and FSR indicating the fraction of optimum homology residues that are strongly hydrophobic (V, I, L, F, M, Y, W) and small (A, G). The standard 1363-based learning set and a 20% threshold was utilized.^†

Variables	β	Standard error	exp(β)	z value	P (>|z|)
Intercept	−0.528	0.031	0.590	−17.247	<0.001
E20	0.342	0.012	1.407	29.162	<0.001
E6	0.862	0.017	2.369	51.156	<0.001
FSR	−0.922	0.031	0.398	−29.690	<0.001
FSHP	−1.646	0.030	0.193	−54.898	<0.001
ALA	−0.267	0.034	0.766	−7.861	<0.001
ARG	0.765	0.034	2.149	22.363	<0.001
ASN	0.358	0.035	1.430	10.236	<0.001
ASP	0.774	0.034	2.168	22.670	<0.001
CYS	−1.543	0.052	0.214	−29.405	<0.001
GLN	0.366	0.036	1.442	10.234	<0.001
GLU	0.985	0.034	2.677	29.047	<0.001
GLY	0.829	0.038	2.292	21.588	<0.001
HIS	−0.114	0.038	0.893	−3.003	0.003
ILE	−0.036	0.027	0.965	−1.353	0.176
LEU	0.202	0.023	1.224	8.767	<0.001
LYS	1.509	0.036	4.522	41.384	<0.001
MET	0.269	0.036	1.308	7.424	<0.001
PHE	0.040	0.030	1.041	1.359	0.174
PRO	0.449	0.034	1.567	13.072	<0.001
SER	−0.166	0.032	0.847	−5.113	<0.001
THR	−0.168	0.032	0.845	−5.291	<0.001
TRP	0.567	0.041	1.763	13.784	<0.001
TYR	0.690	0.029	1.995	24.195	<0.001

^†Note descriptor values for nine PDB chains (1G291, 1L2WA, 1MUWA, 1W85I, 1XC3B, 1XVHA, 2I6CA, 2PI2E) from the original 1363 set are insufficient and here considered null.