Fig 1.
TIA-1 structure and experimental SAXS and SANS data.
(A) All-atom model of TIA-1 and the corresponding coarse-grained Martini model. (B) Experimental SAXS (black) and SANS data (green: 0% D2O, red: 42% D2O, blue: 70% D2O). (C) Corresponding pair distance distribution functions, p(r).
Fig 2.
Results from reweighting the simulation with original force field parameters.
(A) Fit to SAXS data, before (black) and after reweighting at θ = 150 (cyan), θ = 500 (green), and θ = 5000 (purple). (B) vs. ϕeff for selection of θ. (C) Calculated Rg during the simulation, including mean value (black), experimental Rg from SAXS (red) and mean Rg from the reweighted ensembles (green). Corresponding histograms in the right panel. (D) Calculated D13 before and after reweighting.
Fig 3.
Tuning the protein-water interaction strength of the Martini force field.
We varied the protein-water interaction strength by a factor λ, and calculated for each dataset (A-D). Vertical grey lines are the values of λ giving the best fit to the given dataset. No vertical line is given for SANS at 42% D2O as the variation is
is very small. (E) Average Rg as calculated directly from the simulation. The horizontal red line is the value of Rg determined from the SAXS data.
Fig 4.
Results from reweighting the simulation with the optimized force field (λ = 1.06).
(A) Fit to SAXS data with adjusted force fields before (black) and after reweighting at θ = 500 (green). (B) vs. ϕeff for selection of θ. (C) Rg calculated from structures during the simulation, including the mean value (black), experimental Rg from SAXS (red), and mean Rg from the reweighted ensemble (green). Corresponding histograms in the right panel. (D) Calculated D13 before and after reweighting.
Fig 5.
Results from reweighting the simulation with underestimated protein-water interaction strength (λ = 1.04).
(A) Fit to SAXS data with adjusted force field before (black) and after reweighting at θ = 500 (green). (B) vs. ϕeff for a selection of θ. (C) Rg calculated from structures during the simulation, including the mean value (black), experimental Rg from SAXS (red), and mean Rg from the reweighted ensemble (green). Corresponding histograms in the right panel. (D) Calculated D13 before and after reweighting.
Fig 6.
Distributions of Rg (A) and D13 (B) after reweighting.
Reweighted from simulations using λ = 1.00 (black), 1.04 (blue), 1.06 (green), 1.08 (brown) and 1.10 (red). Experimental Rg given in (A) as a dotted line.
Fig 7.
Distributions of Rg (A) and D13(B) after reweighting from unaltered simulation (λ = 1.00) using SAXS and SANS data.
(A) Distribution of Rg and (B) distribution of D13 after reweighting with SAXS alone (red), with SANS at 70% D2O (black), SANS at 42% D2O (green), SANS at 70% D2O (blue) and with all SAXS and SANS data (yellow).
Fig 8.
Cross-validation with SANS data.
Simulations with either (A–D) λ = 1.00 or (E–H) λ = 1.06 were reweighted against SAXS data at several different values of θ. Agreement with (A, E) SAXS data after fitting to the SAXS data (red), or cross-validated with SANS at (B, F) 0% D2O (black), (C, G) 42% D2O (green), (D, H) 70% D2O (blue).
Fig 9.
Theoretical SAXS and SANS scattering for reweighted ensembles in Fig 6.
Distribution 1 is the reweighted ensemble from the simulation with λ = 1.06 (green), distribution 2 is from reweighted ensemble from the λ = 1.04 simulation (blue), and distribution 3 is from the λ = 1.08 simulation (brown). (A) SAXS data. (B-D) SANS calculated with respectively RRM1, RRM2, or RRM3 matched out by perdeuteration to 69% and measured in 100% D2O. Residuals show the relative difference to the scattering from the distribution reweighted from the simulation with λ = 1.06.