A single-model quality assessment method for poor quality protein structure

BMC Bioinformatics

Table 1 Comparison results of various QA methods from Stage 1

QA methods	Corr. on stage 1	Top 1 GDT-TS on stage 1 / Z-score sum	Best model GDT-TS on stage 1 / Z-score sum
Ours (Random Search)	0.42	27.02/59.62	29.43/89.74
Ours (Linear Regression)	0.42	27.42/59.62	29.18/89.74
Ours (Random Forest)	0.32	22.66/14.27	26.65/58.09
Ours (Multilayer Perceptron)	0.37	25.24/45.25	29.15/85.34
DOPE	0.21	23.06/22.11	26.53/51.61
GOAP	0.21	23.48/23.79	27.09/63.87
ProQ4	0.22	23.45/22.58	27.38/61.14
ProQ3D	0.22	23.88/33.54	26.87/64.30
DeepQA	0.21	22.65/14.88	26.07/51.27

Z-score is calculated from the GDT-TS of the selected model and the GDT-TS of all models of the target. Sum the Z scores of each target to get the Z-score sum

ISSN: 1471-2105