Details of the MD simulation setup for each of the systems studied in this work can be found in SI Appendix, Table S16. All systems were simulated using the following force fields: a99SB*-ILDN (11 (link), 12 (link)) with TIP3P (13 ), C22* (14 (link)) with TIP3P-CHARMM (34 (link)), C36m (6 (link)), a03ws (containing modified TIP4P/2005 interactions) (8 (link)), a99SB and TIP4P-Ew (32 (link)) with the Head-Gordon vdW (9 (link)) and dihedral (33 (link)) modifications (termed a99SB-UCB), a99SB-ILDN (12 (link)) with TIP4P-D (7 (link)), and a99SB-disp. (The parameters for the a99SB-disp force field are listed in SI Appendix.) Systems were initially equilibrated at 300 K and 1 bar for 1 ns using the Desmond software (44 ). Production runs at 300 K were performed in the NPT ensemble (45 –47 (link, no link found)) with Anton specialized hardware (48 ) using a 2.5-fs time step and a 1:2 RESPA scheme (49 ). Bonds involving hydrogen atoms were restrained to their equilibrium lengths using the M-SHAKE algorithm (50 ). Nonbonded interactions were truncated at 12 Å, and the Gaussian split Ewald method (51 (link)) with a 32 × 32 × 32 mesh was used for the electrostatic interactions. All simulations were run at 300 K, with the exception of (AAQAA)3, CLN025, and the fast-folding proteins Trp-cage, villin, and GTT, which used simulated tempering (52 ) to improve sampling. In simulated tempering simulations of (AAQAA)3 and CLN025, 20 rungs were spaced geometrically spanning 278–390 K. In simulated tempering simulations of Trp-cage, villin, and GTT, 60 rungs were spaced geometrically spanning 278–400 K.