Screening of selective histone deacetylase inhibitors by proteochemometric modeling

BMC Bioinformatics

Table 2 Goodness-of-fit (R ² ) and predictive ability (Q ² _cv , Q ² _test ) of the obtained 18 models

Model	R²	Q²_cv	Q²_test	Model	R²	Q²_cv	Q²_test
P0-DLI^a	0.9616	0.6564	0.7292	P1-GD-C^b	0.9909	0.6732	0.7191
P0-GD	0.9895	0.6960	0.7331	P2-DLI-C	0.9883	0.6264	0.6519
P1-DLI	0.9619	0.6757	0.7427	P2-GD-C	0.9917	0.5734	0.6215
P1-GD	0.9897	0.7162	0.7542	C(P0,DLI)^c	0.9860	0.6484	0.6941
P2-DLI	0.9614	0.6521	0.7272	C(P0,GD)	0.9914	0.6302	0.6772
P2-GD	0.9894	0.6858	0.7268	C(P1,DLI)	0.9811	0.6686	0.7227
P0-DLI-C	0.9871	0.6462	0.6944	C(P1,GD)	0.9904	0.6625	0.7190
P0-GD-C	0.9916	0.6319	0.6759	C(P2,DLI)	0.9898	0.6505	0.6067
P1-DLI-C	0.9846	0.6626	0.7251	C(P2,GD)	0.9941	0.5967	0.5597

^a P0-DLI means this model is based on protein descriptor P0 and ligand descriptor DLI.
^b P1-GD-C means this model is based on protein descriptor P1, ligand descriptor GD and cross-terms.
^c C(P0,DLI) means this model is based on only cross-terms of P0 and DLI.

ISSN: 1471-2105