Table 2. Protocols used in benchmark calculations.

For main-chain deletion, only the identity scoring matrix was employed. ‘Variable’ thresholds were calculated from the respective sequence alignment. Protocol 1 corresponds to the algorithm published by Schwarzenbacher et al. (2004 ▶).

Protocol	Main-chain window	Threshold	Side-chain matrix	Window	Threshold	B factor
1	0	0.0	Identity	0	1.0	Original
2	3	0.2	Identity	0	1.0	Original
3	5	Variable	Identity	0	1.0	Original
4	0	0.0	BLOSUM62	1	0.2	Original
5	5	Variable	BLOSUM62	1	0.2	Original
6	0	0.0	Identity	0	1.0	Sequence-similarity based (matrix = BLOSUM62, window = 5, factor = 80)
7	5	Variable	Identity	0	1.0	Sequence-similarity based (matrix = BLOSUM62, window = 5, factor = 80)
8	0	0.0	Identity	0	1.0	Accessible surface area-based (factor = 12)
9	5	Variable	Identity	0	1.0	Accessible surface area-based (factor = 12)
10	0	0.0	Identity	0	1.0	Sequence-similarity based (matrix = BLOSUM62, window = 5, factor = 60) + accessible surface area-based (factor = 8)
11	5	Variable	Identity	0	1.0	Sequence-similarity based (matrix = BLOSUM62, window = 5, factor = 60) + accessible surface area-based (factor = 8)
12	5	Variable	BLOSUM62	1	0.2	Sequence-similarity based (matrix = BLOSUM62, window = 5, factor = 60) + accessible surface area-based (factor = 8)