Learning by aggregating experts and filtering novices: a solution to crowdsourcing problems in bioinformatics

BMC Bioinformatics

Table 4 CASP9 comparison on labelled data.

Predictor Name	Institution	ACC	AUC
AEFN		0.801	0.887
GMM-MAPML		0.785	0.874
MAP-ML		0.764	0.859
MV		0.735	0.776
PRDOS2	Tokyo Tech	0.754	0.855
MULTICOM-REFINE	U of Missouri	0.750	0.822
BIOMINE_DR_PDB	U of Alberta	0.741	0.821
GSMETADISORDERMD	IIMCB in Warsaw	0.738	0.816
MASON	George Mason U	0.736	0.743
ZHOU-SPINE-D	Indiana University	0.731	0.832
DISTILL-PUNCH1	UCD Dublin	0.726	0.800
OND-CRF	Umea University	0.706	0.759
UNITED3D	Kitasato University	0.704	0.780
CBRC_POODLE	CBRC	0.694	0.830
MCGUFFIN	University of Reading	0.688	0.817
ISUNSTRUCT	IPR RAS	0.676	0.739
DISOPRED3C	UCL	0.670	0.853
ULG-GIGA	University of Liege	0.588	0.726
MEDOR	Aix-Marseille U	0.579	0.679

Comparisons of AEFN vs. alternative multi-annotator methods (GMM-MAPML, MAP-ML and MV) and individual CASP9 protein disorder predictors.

ISSN: 1471-2105