Feature-based multiple models improve classification of mutation-induced stability changes

BMC Genomics

Table 2 The nine most significant features according to stability selection on the S1676 data set.

Feature	Significance
SIFT score	1.00
Δ hydrophobicity	0.97
relative accessible surface area	0.86
disorder probability	0.81
Δ compressibility	0.80
Δ polarisability	0.58
volume (mean, min, max)	0.54
Δ isoelectric point	0.53
secondary structure probabilities	0.47

Δ refers to the change between the introduced and deleted amino acids; (mean, min, max) was calculated for a window of six neighbouring residues; significance denotes the probability of being selected with stability selection

ISSN: 1471-2164