TABLE I.

Simulated properties for the Aβ42 and Aβ43 peptides for each sampling method-force field combination. Mean and standard deviation averaged over two independent trajectories, except for the 0.8 µs TREx simulation, where block averaging over two 250 ns blocks was used. We noted a systematic shift in J-couplings between the two experimental datasets from Conicella and Fawzi⁴ and Roche et al.; we have applied a −0.4 Hz shift to the Aβ43 J_i values from Conicella and Fawzi before comparison which would bring the Aβ42 results in line with each other to the simulated J-couplings.

	Sampling method and force field combination
	TREx (0.1 µs)	TREx (0.1 µs)	TREx (0.8 µs)	TCW (0.2 µs)	TCW (0.2 µs)	TCW (0.2 µs)
	+ TIP4P-Ew	+ CHARMM36m	+ Amberff99SB	+ Amberff99SB	+ CHARMM36m	+ Amber99SB-ILDN
Peptide	+ Amberff99SB	+ CHARMM-TIP3P	+ TIP4P-Ew	+ TIP4P-Ew	+ CHARMM-TIP3P	+ TIP4P-D
	χ2 between simulated and experimental4,5 J-coupling constants
Aβ42	3.70	4.65	3.70	2.70	3.01	2.89
Aβ43	4.75	3.76	3.65	2.47	2.96	2.71
	EISD score
Aβ42	39.479	3.858	40.014	57.833	47.741	54.937
Aβ43	23.234	25.090	36.530	47.488	39.328	45.338
	Mean and standard deviation of the end-to-end-distance, ⟨Ree⟩ (in Å)
Aβ42	24.3 $\pm$ 0.6	39.3 $\pm$ 3.0	20.5 $\pm$ 1.7	28.4 $\pm$ 0.9	36.8 $\pm$ 1.3	33.5 $\pm$ 7.9
Aβ43	26.5 $\pm$ 3.5	44.4 $\pm$ 1.1	20.3 $\pm$ 0.7	29.1 $\pm$ 2.0	38.2 $\pm$ 1.3	31.9 $\pm$ 0.9
	Mean and standard deviation of FRET efficiencies, ⟨EFRET⟩
Aβ42	0.93 $\pm$ 0.002	0.64 $\pm$ 0.084	0.96 $\pm$ 0.007	0.88 $\pm$ 0.009	0.71 $\pm$ 0.031	0.77 $\pm$ 0.150
Aβ43	0.92 $\pm$ 0.049	0.56 $\pm$ 0.001	0.97 $\pm$ 0.007	0.87 $\pm$ 0.034	0.69 $\pm$ 0.020	0.82 $\pm$ 0.004
	Mean and standard deviation of the radius of gyration, ⟨Rg⟩ (in Å)
Aβ42	12.0 $\pm$ 0.3	17.2 $\pm$ 2.2	11.3 $\pm$ 0.0	12.9 $\pm$ 0.1	15.9 $\pm$ 0.5	14.9 $\pm$ 1.6
Aβ43	11.8 $\pm$ 0.6	17.8 $\pm$ 0.4	11.8 $\pm$ 0.1	13.2 $\pm$ 0.4	16.4 $\pm$ 0.2	15.7 $\pm$ 2.0