GODoc: high-throughput protein function prediction using novel k-nearest-neighbor and voting algorithms

BMC Bioinformatics

Table 1 Dynamic-KNN coverage in partial model with respect to different distance thresholds and voting weight schemes in the cross-validation validation data set. # of seqs: total number of proteins in the set. Distance: distance threshold used in Dynamic-KNN. # of preds: number of predicted proteins and its corresponding proportion in %

Type	Dataset	# of seqs	Distance	Inverse		FunOverlap
Type	Dataset	# of seqs	Distance	# of preds	%	# of preds	%
BPO	CAFA2-Swiss	8146	Q1	2046	25.12	1882	23.10
			Q2	4095	50.27	3654	44.86
			Q3	6112	75.03	5167	64.43
	CAFA3-Swiss	10,163	Q1	2562	25.21	2309	22.72
			Q2	5095	50.13	4470	43.98
			Q3	7601	74.79	6333	62.32
CCO	CAFA2-Swiss	8114	Q1	2039	25.13	1855	22.86
			Q2	4034	49.72	3540	43.63
			Q3	6042	74.46	4898	60.36
	CAFA3-Swiss	9866	Q1	2548	24.91	2204	22.34
			Q2	4922	49.89	4261	43.19
			Q3	7357	74.57	5912	59.92
MFO	CAFA2-Swiss	5211	Q1	1291	24.77	1204	23.11
			Q2	2593	49.76	2366	45.40
			Q3	3902	74.88	3405	65.35
	CAFA3-Swiss	7017	Q1	1756	25.02	1630	23.23
			Q2	3518	50.14	3185	45.39
			Q3	5278	75.22	4573	65.17

ISSN: 1471-2105