Finding minimum gene subsets with heuristic breadth-first search algorithm for robust tumor classification

Wang, Shu-Lin; Li, Xue-Ling; Fang, Jianwen

doi:10.1186/1471-2105-13-178

BMC Bioinformatics

Table 7 The comparison of prediction accuracies by HBSA-KNN, PAM and ClaNC on independent test set

From: Finding minimum gene subsets with heuristic breadth-first search algorithm for robust tumor classification

Methods	Dataset	Number of the top-ranked genes
Methods	Dataset	2	3	4	5	6	7	8	20	40	60
HBSA-KNN	Leukemia (Acc)	84.62	94.23	92.31	94.23	82.69	82.69	82.69	88.46	90.38	92.31
	sensitivity	100	100	100	100	100	100	100	100	100	100
	specificity	71.43	89.29	85.71	89.29	67.86	67.86	67.86	78.57	82.14	85.71
	PPV	75	88.89	85.71	88.89	72.73	72.73	72.73	80	82.76	85.71
	NPV	100	100	100	100	100	100	100	100	100	100
	DLBCL (Acc)	95.24	100	80.95	85.71	80.95	85.71	85.71	85.71	85.71	90.48
	sensitivity	100	100	85.71	85.71	78.57	85.71	85.71	85.71	85.71	92.86
	specificity	85.71	100	71.43	85.71	85.71	85.71	85.71	85.71	85.71	85.71
	PPV	93.33	100	85.71	92.31	91.67	92.31	92.31	92.31	92.31	92.86
	NPV	93.33	100	85.71	92.31	91.67	92.31	92.31	92.31	92.31	92.86
	Prostate (Acc)	88.24	76.47	82.35	85.29	82.35	79.41	76.47	76.47	82.35	85.29
	Sensitivity	100	100	100	100	100	88.89	88.89	88.89	88.89	100
	Specificity	84	68	76	80	76	76	72	72	80	80
	PPV	69.23	52.94	60	64.29	60	57.14	53.33	53.33	61.54	64.29
	NPV	100	100	100	100	100	95	94.74	94.74	95.24	100
	SRBCT	75	95	95	100	95	95	95	95	95	100
	ALL	75	82	87	94	92	93	93	96	97	99
	Colon (Acc)	65	70	80	75	80	80	75	75	75	75
	sensitivity	75	83.33	91.67	91.67	91.67	83.33	83.33	83.33	75	75
	specificity	50	50	62.50	50	62.50	75	62.50	62.50	75	75
	PPV	69.23	71.43	78.57	73.33	78.57	83.33	76.92	76.92	81.82	81.82
	NPV	57.14	66.67	83.33	80	83.33	75	71.43	71.43	66.67	66.67
PAM	Dataset	Number of the selected genes
	Dataset	2	4	6	8	10	12	16	20	40	60
	Leukemia	82.69	90.38	90.38	90.38	92.31	94.23	96.15	96.15	98.08	98.08
	DLBCL	66.67	66.67	66.67	66.67	66.67	66.67	66.67	66.67	66.67	66.67
	Prostate	73.53	73.53	73.53	73.53	73.53	73.53	73.53	73.53	73.53	73.53
	SRBCT	45	45	75	75	85	95	95	95	95	95
	ALL	43	61	61	68	68	83	85	85	86	86
	Colon	65	75	70	70	70	75	75	75	75	75
ClaNC	Dataset	Number of the selected genes
	Dataset	1 × k *	2 × k	3 × k	4 × k	5 × k	6 × k	7 × k	8 × k	9 × k	10 × k
	Leukemia	86.54	90.39	90.39	92.31	90.39	94.23	94.23	94.23	94.23	96.15
	DLBCL	80.95	95.24	95.24	95.24	95.24	80.95	76.19	71.43	71.43	71.43
	Prostate	73.53	85.29	79.41	76.47	76.47	79.41	79.41	76.47	76.47	79.41
	SRBCT	85	95	95	95	95	95	95	95	95	95
	ALL	86	95	97	99	98	98	99	99	99	98
	Colon	65	65	65	70	70	75	75	75	75	75

* k denotes the number of subclasses in each dataset, which ranges from two to six. For example, for ALL dataset, the size of the gene subset selected ranges from six (1 × 6) to sixty (10 × 6).

Back to article page

ISSN: 1471-2105

Contact us

General enquiries: journalsubmissions@springernature.com