Table 2. Experimentally Measured and Computationally Predicted NOE Distances (R) for Some Cross-Peaks^a.

	EXP	1.9 μs	15 × 50 ns
	Rexp (Å)	RA99χ (Å)	RA99χ (Å)	RA99 (Å)
G1H2′-A2H8	2.8	2.8 (2.4/5.6/3.8)	2.3	3.6
A2H2′-C3H6	2.6	2.7 (2.5/2.6/3.1)	2.4	3.3
A2H3′-C3H6	2.7	2.6 (2.6/2.3/3.0)	2.5	2.7
A2H2–C3H1′	3.9	3.4 (3.3/3.2/3.7)	3.2	5.6
C3H2′-C4H6	2.8	3.1 (2.7/4.9/3.3)	2.7	3.2
C3H3′-C4H6	2.8	2.8 (2.5/4.4/3.0)	2.5	3.0
rmsdb		0.24	0.39	0.84

^aR is the “distance” calculated from measured or predicted NOE (1/R⁶). If more than one conformation is present, then R does not represent any physical distance. The exp, A99χ and A99 subscripts stand for experiment, AMBER99χ, and AMBER99. Predictions are from single 1.9 μs MD simulation of AMBER99χ and for combined fifteen 50 ns MD of both force fields (15 × 50 ns) using structures extracted every 0.5 ps (see Table S7 for detail). “Distances” in parentheses are for the following time domains of the 1.9 μs simulation: 1–700 ns/701–1100 ns/1101–1900 ns.

^brms deviation from experimental results. “Distances”, R, for AMBER99χ simulations of 50 ns starting from A-form are consistently shorter than experimental “distances” from NMR, presumably because GACC is not entirely A-form in solution, but remains A-form in 50 ns simulations (see comparisons of NOEs in Table S11). The shorter “distances”, however, fall within the range consistent with NOE distance restraints (Table 1). The 1.9 μs simulation allows more time for sampling structures, so that the systematic effect of short simulation time is reduced (see Figure S16). This results in a lower rmsd relative to the experimental “distances”.