The H-factor as a novel quality metric for homology modeling

Table 1 Comparison between the H-factor, cRMS, DOPE and QMEAN scores

CASP targets	Scoring function (a)				H-factor (%)	cRMS (Å) (b)	% ID (c)	DOPE (d)	QMEANnorm (e)
	(1)	(2)	(3)	(4)
T0295	1.0	1.9	1.3	3.8	19	1.67	46	-33940	0.735
T0295* (f)	1.0	1.9	1.9	3.8	21	1.71	46	-24317	0.196
T0522	2.0	6.2	1.5	5.0	37	2.19	38	-14185	0.771
T0522* (f)	2.0	6.9	2.0	5.2	40	2.71	38	-12687	0.681
T0521	4.2	7.5	3.4	5.4	51	4.09	24	-18484	0.695
T0521^# (g)	6.1	7.6	3.6	6.8	60	4.24	23	-15968	0.508
T0544	7.0	8.3	4.0	8.3	69	6.11	17	-10652	0.192

(a) The four scoring functions included in the H-factor measure the quality of the secondary structure prediction (score (1)), the diversity of the sequence alignment (score (2)), the structural diversity of the models generated (score (3)), and the similarity of the predicted structures for the functional domains in the target, compared to the structures of the same domains found in the PDB (score (4)).
(b) Average cRMS (over Cα) between the 10 different models generated and the actual experimental structure for the target.
(c) Sequence identity between the target sequence and the template sequence.
(d) DOPE scores from MODELLER [24].
(e) QMEAN normalized score [25].
(f) T0295^* and T0522* are experiments in which the sequence alignment between the target and its framework has been deliberately modified with a shift of a single amino acid in the sequence alignment (see text for details).
(g) T0521^# is an experiment in which a less suitable template has been deliberately chosen compared to T0521 (see text for details).