Table 5.

Quartile analysis of dataset protein properties vs. SVM scoring.

	Original n	Subset n	%	μH	Q	NK/NK+NR	H	PI	σ
Category	Total	μH*Q > 1.0							SVM
Known bacteriocins	308	43	14	0.52	3.2	0.68	0.38	8.68	0.90
Bacteriocin-related	15	9	60	0.56	3.4	0.91	0.29	9.98	0.64
Non-bacteriocin	52	10	19	0.52	2.7	0.27	0.34	8.18	0.72
Uncharacterized	1,038	85	8	0.53	3.2	0.56	0.33	8.67	0.91
Category	Total	μH*Q > 0.50							SVM
Known bacteriocins	308	79	26	0.46	2.6	0.69	0.42	7.9	0.80
Bacteriocin-related	15	10	66	0.57	3.2	0.92	0.32	9.6	0.63
Non-bacteriocin	52	15	29	0.50	2.4	0.38	0.35	8.1	0.65
Uncharacterized	1,038	208	20	0.49	2.3	0.63	0.36	7.6	0.80
Category	Total	μH*Q > 0.25							SVM
Bacteriocins	308	161	52	0.36	2.1	0.75	0.42	7.2	0.58
Bacteriocin-related	15	12	80	0.52	2.8	0.76	0.35	9.1	0.54
Non-bacteriocin	52	16	31	0.50	2.3	0.4	0.35	7.9	0.67
Uncharacterized	1,038	319	31	0.44	1.9	0.65	0.38	7.2	0.65

The μH*Q values represent different percentile cutoffs for peptide groups (dark orange, >1.0; middle orange; >0.50; and light orange, >0.25). Definition legend: μH–hydrophobic moment; Q–charge; N_K/N_K+N_R relative percentage of lysine vs. arginine; H–hydrophobicity; PI–isoelectric point.