PowerFit is a tool for rigid body fitting of atomic structures in cryo-electron density maps, outputting reliability indicators for the success of fitting subunits in density maps, through the use of the Fisher z-transformation and associated confidence intervals.
In most cases the default parameters of PowerFit will be reasonable and don’t have to be changed. In cases where the fitting results are unsatisfactory, improvements might be achieved by disabling the Laplace pre-filter, disabling the core-weighted correlation, or increasing the rotational sampling. Increasing rotational sampling by a factor of 2 results in approximately 8 times more rotations sampled and thus a significant increase in the runtime of a Job.
A short Tutorial on the use of the PowerFit web server to fit the methyltransferase KsgA (4ADV) into a cryo EM map of the complex with the 30S maturing E. coli ribosome (EMD-2017) is available at bonvinlab.org.
The PowerFit tutorial demonstrates the use of the freely available PowerFit software.
Cryo-EM map:
A file in any of the common format for Cryo-EM maps (mrc, ccp4, or map) in which the model will be fit.
Map resolution:
Resolution of the Cryo-EM map Mandatory!
Atomic structure:
A file in pdb or mmCIF format containing the coordinates of the model that will be fit into the density map.
A custom tag to identify your run. Supported characters: a-z A-Z 0-9 _ + - = / , . :
Rotational sampling interval - Rotational sampling density in degree. Increasing this number by a factor of 2 results in approximately 8 times more rotations sampled.
Chain ID(s) - The chain IDs of the structure to be fitted. Multiple chains can be selected using a comma separated list, i.e. A,B,C. Default is the whole structure.
Laplace pre-filter - Use the Laplace pre-filter to filter the density data. Can be combined with the core-weighted local cross-correlation.
Core-weighted scoring function - Use a core-weighted local cross-correlation score. Can be combined with the Laplace pre-filter.
Number of top solutions - Number of Top solutions for which the model in the cordinate frame of the density map will be returned. This number will be capped if less solutions are found as requested.
Density map resampling - Resample the density map prior to fitting.
Resamplig rate - Adjust resampling of the density map to a specific factor of the Nyquist rate.
Density map trimming - Trim the density map to a predefined intensity cutoff.
Trimming cutoff - Intensity cutoff to which the map will be trimmed. Default is 10 percent of maximum intensity.
Result Page
The results page provides links to the output files of a submitted PowerFit run and an overview of the results.
The solutions table provides a summary over the best 15 non-redundant solutions found by PowerFit as reported in the solutions.out file (see below). For each solution the table provides the rank, the cross correlation score, the Fisher z-score, the error-normalized z-score (z-score/σ), and the sigma difference to the best fit.
In a previous investigation we fitted 379 individual chains in 6 ribosome cryo-EM density maps starting at a resolution of 6Å up to 30Å. Successful fits were obtained in >99% of the cases for which the difference in error-normalized z-score between the two top scoring solutions was larger than ~2.5. Conversely, in less than 3% of the failed cases did the difference exceed one. (Van Zundert and Bonvin 2016)
The true-positive rate is given versus the difference in Fisher z-score standard deviations between the top 2 solutions. The fitting results were binned in 6 bins, starting from 0 to 3 sigma with a step size of 0.5.
To enhance the interpretation of the results, the entries in the table are colored in a green gradient up to a sigma difference of 3. Coloring is however only applied if the sigma difference to Fit 10 is below 3 in order to avoid higlighting runs with no distinction between the fits.
Note that in case of symmetrical complexes multiple equally well scoring fits might be reported.
Fit [N]:
Next to the summary table, a detailed entry for the top N fits (see Number of top solutions) is provided including its rank, the cross correlation score, the Fisher z-score, the error-normalized z-score (z-score/σ), the sigma difference to the best fit, and a PDB file of the target complex representing the fit in the coordinate frame of the provided map.
Once the image generation (decoupled from the main job) with UCSF Chimera is completed, images for each fit are displayed in six different views (Intensity cutoff of the map is set to mean+2σ).
Visualizing models fitted in the density map
While the autogenerated images on the result page give a first impression of the fit, models fitted in the density map can be visualized in more detail by opening the input density map (target.map) used in the powerfit run together with the different models output in the result directory (fit_N.pdb) with UCSF Chimera.
Downloadable Output
Submitting a PowerFit run generates several files available for Download:
Files generated by PowerFit :
PDB files of the top N best fits (see Number of top solutions).
A plain text file with all the non-redundant solutions found, ordered by their correlation score. The first column shows the rank, column 2 the correlation score, column 3 the Fisher z-score, column 4 the zscore as factor of standard deviations (z/σ (see N. Volkmann 2009, and Van Zundert and Bonvin 2016); column 5 to 7 are the x, y and z coordinate of the center of the chain; column 8 to 17 are the rotation matrix values.
A cross-correlation map, showing at each grid position the highest correlation score found during the rotational search.
A log file of the PowerFit run, including the input parameters with date and timing information.
Files generated by the web server:
A text file in json format containing the key parameters provided at the input form.
Image of Fit N within the provided density map in one six different views.
If you have further questions or run into issues submitting your data, please contact us at powerfit.server[at]gmail.com.