Running Foldhunter in Chimera

Foldhunter localizes any probe structure (PDB or MRC file) to a target density (MRC file only).  To increase the speed of foldhunter, both the probe and target maps are shrunk and the optimal angular search is calculated.  Options for identifying unique peaks (heirarchical search), refinement of selected chains/fragmentd and the standard SMART foldhunter are also included.

1. INPUTS

Required Parameters:
Target file Any even, cubic MRC file
Probe File Any PDB or MRC file
Apix Angstroms/pixel
Res Resolution of the target Map
NOTE: Both the probe file and the target file can be entered through the selection button, which displays all appropriate open models in chimera. Additionally, the user can specify a file not currently open in Chimera by selecting the "use external file" option button and typing the name of the file, in the current working directory, into the entry box.

Optional Parameters:
Shrink
 Specifies the amount of scaling done to the probe and target maps. Shrinking reduces the size of the amount and thereby decreases the amount of time required to do an exhaustive search. After identifying the top localization, a refinement of the fit is done through a limited angular and distance search using the unshrunk maps. default=2
Keep
 Number of peaks kept in the log file. deafult=20
Filters Pre-processing filters for the target map. default=None
Binary
Sets everything below a value to 0, above the value=1
Threshold
Sets everything below a value to 0
Normalize
Standard proc3d normalization
Laplcaian
Standard proc3d laplacian filter

SMART
 Smart refinement in foldhunter. Runs foldhunter with a fixed angular search (da=30) and then refines top peak to below da=1 through several iterations. default=disabled
Heirarchical Search Allows the user to look for multiple unique peaks in the first, shrunk run of foldhunter and subsequently refine these peaks. default=disabled
Peak distance
 Distance, in pixels, to be considered unique solution. deault=1
Depth
Number of unique peaks to examine. Will only consider those peaks reported in logfile as defined by keep variable. deafult=1
Results
User selectable list of possible unique fits generated from heirarchical search
Open
Opens selected file for viewing
Refinement: Allows the user to select a portion of an x-ray structure or an entire MRC file for further refining the foldhunter fit (da=1) into the target structure. deafult=disabled
Refine model
 User selectable list of possible models used for refinement
Angle
Angular range to search over. default=1
Distance
Distaance in pixels to perform refinement. default=1


2. Control Buttons:        

Apply Executes foldhunter with desired inputs and options within Chimera
Close Exits foldhunter submenu
Help Launch this help window

 
3. Outputs

fh-*.mrc Transformed MRC file (smart, refine, shrink, solution number)
fh-*.pdb Transformed PDB file (smart, refine, shrink, solution number)
log-*.txt Text log generated from foldhunter (smart, refine, shrink, solution number)


4. FAQ

1. What inputs will foldhunter accept?
For the target files, foldhunter accepts all volume files read by EMAN (mrc, imagic, etc...) although Chimera only reads MRC files directly.  Probe files may be in either PDB format or one of the aforementioned supported volume files.

2. What is the output of foldhunter?
Foldhunter will produce a transformed PDB, transformed MRC and a text log file when the probe is a PDB file.  The output PDB file only contains ATOMS and excludes all header information. Only a transformed MRC and text log file will be produced when the probe is a volume file.

3. Why is my EM density not scaled and center correctly?
Chimera and EMAN use different conventions for the assigning the origin of a density map.  In the volume viewer, click on the data button in the top of the panel and set the voxel size to ang/pixel value.  Also set the X,Y,Z origin to (-0.5*MapSize)*(voxel size).

4. I don't want to use the shrink option. How do I disable it?
Set shrink to 1.

5. Do I have to run a complete run of foldhunter before using the refine option?
No.

6. Is there any major difference between this version of foldhunter and the EMAN foldhunterP?
Not really.  The Chimera version of foldhunter uses foldhunterP and wraps it with some additional control logic and new options like shrinking and refinement.

7. Is there a non-graphical version of this new foldhunter?
Yes. EMAN is now distributed with foldhunter.py which mimics much of the behavior found in the Chimera version of foldhunter.

8. How long does foldhunter take to run?
Foldhunter can take as little as a few seconds to as much as a few hours.  Typical interemdiate resolution structures in a 128x128x128 volume will take anywhere from 3-10 minutes.

9. Why does foldhunter take longer at lower resolutions?
The andgular search (da) is set using 1.5*((180*resolution)/(pi*(map size/2))) during the shrink step and 1 during the  in the final step.  DA is set to 30 when using the SMART option and 1 during refinements. However, the angular range to consider in the final foldhunter step and refinement is (180*resolution)/(pi*(map size)). For this reason, a low resolution that is the same size as another higher resolution map will take longer.  It also takes longer time to generate a MRC file from the PDB file at low resolution.

10. Why doesn't my fit look right?
Foldhunter will dock a probe to a target file to the most probable site.  This does not necessarilt have to be the correct site though. It may be possible to achieve the correct through filtering the target file. A threshold or binary filter may help if the target file has a large density value range or has regions that are considerable more dense than others. Normalization may help to put the target EM density on the same relative scale as the probe denssity.  With little noise, the laplacian filter may enhance the edges of the target map.  A seperate bilateral filtering (found in proc3d as blfilt and a forethcoming filter module in chimera)  of the density at a particular threshold may also help to enhance realtively weak or \noisy density.

11. Are there any statistics to judge the qualtiy of fit in foldhunter?
Sort of.  Each fit is given a correlation value.  This value is relative only to the particular probe and target. Also mean and standard deviation are given for the top 1000 fits.  A relative measure of fit can also be obtained by looking at the relative fall-off of the correaltion coefficient with respect to unique peaks. A truelu unique peak should have a rapid fall-off from the bet fit to other possible fits.

12. Are you really as cool as I think?
Even more so.

5. Examples

1. Fit a PDB file to an EM density.
Open chimera and read in the files of interest. Under Tools, EMAN open foldhunter.  Select the MRC file as the target and the PDB file as the target.  Enter A/pix and resolution in the required options.  Push "Apply". Wait for foldhunter to finish execution.  A new, transformed PDB file will appear.  Repeat as necessary. Enjoy.

2. Fit a PDB file into EM density map at several unique locations.
Open chimera and read in the files of interest. Under Tools, EMAN open foldhunter.  Select the MRC file as the target and the PDB file as the target.  Enter A/pix and resolution in the required options.  Select "heirarchical search" and enter the distance in pixels that will specify the unique peaks and the total number of unique peaks to identify.  Give keep a sufficently high number of total fits to log. Push "Apply". Wait for foldhunter to finish execution.  A new, transformed PDB file will appear.  The results selctor can then be used in conjunction with the open button to open all unique peaks found. Use Model Panel to toggle the status or remove the tranformed PDB files. Enjoy.

3. Fit a PDB structure and then refine a chain from the tranformed PDB file
Open chimera and read in the files of interest. Under Tools, EMAN open foldhunter.  Select the MRC file as the target and the PDB file as the target.  Enter A/pix and resolution in the required options.  Push "Apply". Wait for foldhunter to finish execution.  A new, transformed PDB file will appear.  Select the "Refinement" option.  Select the tranformed PDB file in the "refine model" selector.  Select the chain using Chimera's selection tools. Set the angle and distance parameters. Push "Apply". Wait for foldhunter to finish execution.  A new, transformed PDB file will appear.  Repeat as necessary. Enjoy.


6. More Help

Help... I still don't get it./I can't read./I can't follow instructions.
You can contact me or check out the foldhunter/helixhunter webpage and EMAN documentation.

http://ncmi.bcm.tmc.edu/~wjiang/hhfh
http://ncmi.bcm.tmc.edu/homes/stevel/EMAN/doc

contact information:
Matthew Baker
mbaker@bcm.tmc.edu
713-798-6989