PHDcleav: a SVM based method for predicting human Dicer cleavage sites using sequence and secondary structure of miRNA precursors

Table 1 Performance of 'RNAfold' derived sequence-based SVM models for Dicer cleavage sites at 5p and 3p arm.

Features	Window size	CD-5p (sequence-based)				CD-3p (sequence-based)
		Sn	Sp	Ac	Mc	Sn	Sp	Ac	Mc
Mono	8	58.20	61.80	60.00	0.20	58.38	59.28	58.83	0.18
	10	63.78	58.92	61.35	0.23	60.36	61.44	60.90	0.22
	12	61.08	63.96	62.52	0.25	64.68	59.46	62.07	0.24
	14	60.36	64.86	62.61	0.25	63.78	66.49	65.14	0.30
Dinuc	8	59.82	59.82	59.82	0.20	61.80	52.25	57.03	0.14
	10	60.54	58.20	59.37	0.19	58.74	52.97	55.86	0.12
	12	61.44	56.40	58.92	0.18	60.90	61.62	61.26	0.23
	14	60.72	65.05	62.88	0.26	60.36	65.95	63.15	0.26
Trinuc	8	60.90	57.48	59.19	0.18	58.92	56.04	57.48	0.15
	10	55.14	61.62	58.38	0.17	59.82	53.15	56.49	0.13
	12	58.92	55.14	57.03	0.14	63.24	57.12	60.18	0.20
	14	60.72	58.74	59.73	0.19	64.68	60.90	62.79	0.26
Binary	8	60.54	61.62	61.08	0.22	65.23	65.05	65.14	0.30
	10	61.26	63.78	62.52	0.25	67.75	63.06	65.41	0.31
	12	67.57	65.05	66.31	0.33	69.55	66.13	67.84	0.36
	14	71.89	77.12	74.50	0.49	70.99	63.96	67.48	0.35

Models were developed using composition and binary pattern of different window sizes. Patterns were taken from sequence of Dicer cleavage site of 5p arm (sequence of CP-5p) and 3p arm (sequence of CP-3p) using miRNA.str data.
CD-5p: Dicer cleavage site at 5p arm, CD-3p: Dicer cleavage site at 3p arm, Mono: Mononucleotide, Dinuc: Dinucleotide, Trinuc: Trinucleotide, Binary: Binary pattern, Sn: sensitivity, Sp: specificity, Ac: accuracy, Mc: Matthews correlation coefficient.

ISSN: 1471-2105