All-Atom Four-Body Knowledge-Based Statistical Potentials to Distinguish Native Protein Structures from Nonnative Folds

Table 1

Summary data for the protein structure training set (1417 single chains).

Atomic alphabet	Count	Proportion
C	1572222	0.634149
N	425874	0.171774
O	469869	0.189520
S	11299	0.004557
Total atom count	2479264
Total tetrahedron counts
No edge-length cutoff	16152638
12 Å edge-length cutoff	15497203
8 Å edge-length cutoff	14567713
4.8 Å edge-length cutoff	9569503

Same counts regardless of atomic alphabet size.