Engineering proteinase K using machine learning and synthetic genes

BMC Biotechnology

Table 2 Vector weights calculated for amino acid substitutions.

	RR		Lasso		PLSR		SVMR
Substitution	M	σ	M	σ	M	σ	M	σ
S107D	-0.03	0.13	0.00	0.02	-0.70	0.26	-0.16	0.13
S123A	-1.00	0.13	-0.41	0.35	-1.42	0.23	-0.93	0.14
I132V	0.04	0.44	0.04	0.55	0.32	0.76	-0.34	0.29
M145F	-1.46	0.19	-2.27	0.49	-1.98	0.32	-1.58	0.20
Y151A	1.18	0.23	0.91	0.23	1.66	0.37	0.91	0.15
V167I	-0.97	0.13	-1.09	0.15	-1.10	0.17	-0.79	0.14
L180I	-0.23	0.15	-0.05	0.10	-0.35	0.19	-0.36	0.13
Y194S	0.27	0.20	0.00	0.01	0.94	0.73	0.01	0.14
A199S	-1.16	0.39	-1.09	0.46	-2.66	0.98	-0.86	0.21
K208H	0.28	0.15	0.07	0.12	0.52	0.18	0.36	0.17
R237N	-0.93	0.09	-0.91	0.13	-1.21	0.12	-0.86	0.15
V267I	-0.48	0.11	-0.32	0.14	-0.68	0.13	-0.16	0.12
S273T	0.12	0.14	0.01	0.06	0.28	0.19	-0.05	0.17
G293A	1.95	0.13	2.24	0.14	2.10	0.17	1.70	0.13
K332R	0.07	0.13	-0.01	0.05	0.02	0.14	0.09	0.15
S337N	-0.02	0.14	0.03	0.09	-0.20	0.15	0.03	0.14
P355S	-1.08	0.12	-1.20	0.15	-1.25	0.13	-1.10	0.15
	LPSVMR		LPBoostR		MR		ORMR
Substitution	M	σ	M	σ	M	σ	M	σ
S107D	-0.01	0.20	-0.02	0.21	-0.35	0.24	-0.35	0.24
S123A	-0.41	0.43	-0.40	0.41	0.52	0.93	0.52	0.93
I132V	0.22	0.69	0.18	0.56	2.61	0.91	2.61	0.91
M145F	-2.39	0.53	-2.39	0.53	-5.33	1.05	-5.33	1.05
Y151A	0.82	0.24	0.82	0.25	0.64	0.24	0.64	0.24
V167I	-0.99	0.24	-0.98	0.23	-1.63	0.24	-1.63	0.24
L180I	-0.20	0.16	-0.19	0.16	0.60	0.23	0.60	0.23
Y194S	-0.02	0.08	0.21	0.38	4.59	2.10	4.59	2.10
A199S	-0.49	0.33	-0.73	0.54	-4.92	2.03	-4.92	2.03
K208H	0.16	0.18	0.14	0.18	0.01	0.13	0.01	0.13
R237N	-0.96	0.29	-0.96	0.29	-1.59	0.57	-1.59	0.57
V267I	-0.41	0.23	-0.41	0.23	-1.33	0.14	-1.33	0.14
S273T	0.19	0.33	0.15	0.28	0.96	0.58	0.96	0.58
G293A	2.18	0.25	2.20	0.25	3.20	0.14	3.20	0.14
K332R	0.12	0.19	0.14	0.21	-0.33	0.13	-0.33	0.13
S337N	0.27	0.28	0.26	0.28	0.34	0.59	0.34	0.59
P355S	-1.34	0.35	-1.35	0.34	-1.95	0.57	-1.95	0.57

Mean (M) and standard deviation (σ) values are shown for the 19 substitutions for which weights were calculated using machine learning. The values were calculated from 1000 subsamples of the variants with measurable activity from sets 1 and 2, where 5 variant sequences were randomly omitted from each subsample.

ISSN: 1472-6750