Final Models Selection Using WAXSiS

Instructions for finalizing and selecting the best models in SAXS-A-FOLD.

Overview

The Final Models Selection Using WAXSiS tab is the final step in SAXS-A-FOLD, allowing the computation of I(q) profiles for NNLS-selected models and optional adjacent frames from the original MMC run. This ensures a comprehensive analysis and selection of the best models for SAXS data interpretation.

Fields in the Final Models Tab

Besides the standard Project Name, Description, and Downloads fields, this tab includes the following:

  • Additional Adjacent Frames Count: Specifies the number of structures flanking the NNLS-selected models to include in the computation. Default value: 0.
    • This feature captures potentially overlooked models that may better contribute to the NNLS fitting procedure.
    • Using this option significantly increases the computation time, with progress monitored via a progress bar and associated messages.
  • WAXSiS convergence mode [details]: Two options are available from a pull-down menu: Quick and Normal. Default: Normal.
    • Quick convergence is useful to get a first final set of models more quickly.
    • Normal convergence is recommended, but using this option significantly increases the computation time.

Steps to Finalize Model Selection

  1. Set Additional Adjacent Frames: Adjust the Additional Adjacent Frames Count field if needed. Use caution, as increasing this count will prolong computation time.
  2. Select WAXSiS Convergence mode: Select between Quick and Normal. Be aware that Normal will require roughly twice the time to complete.
  3. Start Computations: Press Submit to compute I(q) profiles for NNLS-selected models and any additional frames. Progress is tracked in real time, and the projected computation time is updated as more runs are completed.

Outputs and Visualization

Once the I(q) calculations are completed, a final NNLS fit is automatically performed. The results include:

  • A new plot to the right of the original data plot, showing:
    • WAXSiS-calculated I(q) profiles for each NNLS-selected structure, scaled to the experimental I(q) curve.
    • The reconstructed I(q) curve based on the percent contribution of each computed I(q) curve, including propagated SD.
    Users can toggle individual curves on/off by clicking their corresponding entry in the legend.
  • Below the plot, a list of NNLS-selected models is displayed, along with their percent contribution to the fit.
  • To the left of the NNLS-fit numerical results, two stacked histogram plots, showing:
    • On the top, the % contribution of each NNLS-selected model as a function of their Rg value, color coded as the curves shown in the NNLS fit graph. Three inverted triangle report the Rg values of the starting structure (blue), the weighted average of the selected models (green), and that calculated from the experimentally-derived P(r) (brown). Hovering with the mouse on each bar or triangle will reveal its legend, including the Rg values and for the bars the model's % contribution.
    • Below it, the Rg values frequency of the entire pool and of the sub-selected pool are reproduced.
  • Options to download:
    • The entire dataset of WAXSiS-calculated I(q) profiles, including those shown in the plot.
    • The selected models in a single NMR-style file.

The selected models are visualized in an interactive JSmol viewer:

  • Models are spatially aligned on the previously chosen sequence.
  • Each model is colored as its corresponding curve in the NNLS-fit graph and bar in the histogram plot, and displayed in ribbons mode by default.
  • Users can access additional viewing options, such as showing only a particular model, by right-clicking on the viewer.
Return to Main Help