Skip to content

Analysis

Jump to bottom
Michael A. Cianfrocco edited this page Jun 7, 2018 · 1 revision

Home > Analysis

##Analysis

Below is a list of analysis protocols/routines to help users interpret and analyze 2D images and 3D models:

  1. Aligning 2D class averages with projections of 3D model
  2. Filtering, thresholding, masking 3D models
  3. Docking PDB models into 3D EM maps
  4. Calculating local resolution of 3D models
  5. Difference density maps for 2D or 3D inputs
  6. Measuring dimensions in EMAN, EMAN2, and Chimera

###Aligning 2D class averages with projections of 3D model

An important validation experiment for your 3D model is to test how well projections of the 3D model align to 2D class averages of the same dataset. While not a guarantee that your model is 100% accurate, this test must be satisfied if you are to being to trust your 3D density.

This program will project your 3D model and then align the class averages to projections that have the highest score. Then, the program will output a stack of particles, where it shows each average alongside the best matching projection of the 3D model.

To run the program:

$ EM-processing-master/Analysis/compare_avgs_to_3Dmodel.py 
Usage: compare_avgs_to_3Dmodel.py -i  --num=[number of class averages] --boxsize=[boxsize of data] -v 

Options:
  -h, --help     show this help message and exit
  -i FILE        Stack of 2D class averages in .img format.
  --num=INT      Number of class averages
  --boxsize=INT  Boxsize of particles and 3D model
  -v FILE        3D volume to be aligned to 2D averages (.mrc or .spi format)
  --angstep=INT  Angular step for projecting 3D model (Default=10 degrees)
  -d             debug

Where the angular step is defined as the increment (in degrees) that will be the minimum distance between two different projections. A small angular step (e.g. 1 degree) represents a very fine sampling of the 3D model, whereas a large angular step (e.g. 20 degrees) represents a coarse sampling (e.g. 20 degrees) of the 3D model.

Example command:

$ EM-processing-master/Analysis/compare_avgs_to_3Dmodel.py -i sum7_42_30.img --num=30 --boxsize=128 -v run1_local_ct15_it015_class003_dc2_128.mrc --angstep=5

Output:

This program will output an imagic stack that will be in the following order:

class average #1
best matching projection
class average #2
best matching projection 
...

###Docking PDB models into 3D EM maps

###Calculating local resolution of 3D models

###Difference density maps for 2D or 3D inputs

In order to make a strong conclusion that two images or two 3D models show significantly difference density, you need to calculate a difference map.

A difference map is the result of subtracting normalized 2D or 3D objects from each other, where the resulting difference density is measured in standard deviations away from the average.

By calculating a difference map and THEN observing 2D/3D densities that have pixel values >3.5, this means that the density is above noise (which is from -3.5 to 3.5 intensity values), and is therefore significant.

There is a single python script that can calculate 2D or 3D differences between two, individual 2D images or two 3D models:

$ EM-processing-master/Analysis/calculate_difference_density.py 
Usage: calculate_difference_density.py --input1=[input] --input2=[input] --type=2D/3D

Options:
  -h, --help     show this help message and exit
  --input1=FILE  Image/Volume from which input2 will be subtracted. (2D or 3D
                 input, .spi or .mrc format)
  --input2=FILE  Image/Volume that wil be subtracted from input1 econd input
                 will be subtracted. (2D or 3D input, .spi or .mrc format)
  --type=TYPE    Input type: 2D or 3D.
  -d             debug

This script handles .spi or .mrc inputs, and 2D or 3D data types.

It is important that you are providing individual 2D images, and not a stack with a single particle. You can read more here about this conversion.

Example command:

An example command for calculating a 3D difference map is the following:

$ EM-processing-master/Analysis/calculate_difference_density.py --input1=volume1.mrc --input2=volume2.mrc --type=3D

Output:

The output from this command will be a file named:

{input1 name}_minus_{input2 name}.mrc/spi

This filename shows you what the program calculated: the subtraction of input2 from input1.

Interpreting density maps

For 2D difference maps, you will need to open the resulting difference maps in EMAN or EMAN2, and then measure the intensities of the pixels.

  • For either program, middle click on the image and then select 'Probe'. Then mouse/click on pixels in the image the program will tell you the intensity values.

For 3D difference maps, if you would like to visualize what additional densities are present on input1 volume, you can open input2 volume and the difference map in Chimera. Then, threshold the difference to values above 3.5, and you will be visualizing significantly difference densities.

###Measuring dimensions in EMAN, EMAN2, and Chimera

Home

Clone this wiki locally