Appion

h1. Hierarchical or K-means Clustering

This method clusters particles using k-means or hierarchical ascendancy according to metrics obtained via feature analysis procedures such as correspondence analysis.  As described in _Frank et. al Journal of Microscopy v150,99-115 (1988)_.

h2. General Workflow:

_Note:  "Run Particle Clustering" can be accessed directly from a feature analysis run, or via the "Run Particle Clustering" link in the Appion sidebar menu.  In the latter case, you will be taken to the list of feature analyses that have been completed, where you can select the feature analysis run which you wish to cluster._

# Check that the appropriate alignment and feature analysis runs have been chosen.

# Click on the dendrogram to enlarge.

# Select the boxes beneath eigen images that you wish to use for clustering.

# Write in a list of numbers of classes to calculate.  The default is to calculate 4, 16, and 64 classes.

# Use the radio button to select Hierarchical or K-means clustering.

# Make sure that "Commit to Database" box is checked.  (For test runs in which you do not wish to store results in the database this box can be unchecked).

# Click on "Run Cluster Coran" to submit your job to the cluster.  Alternatively, click on "Just Show Command" to obtain a command that can be pasted into a UNIX shell.

!http://emg.nysbc.org/attachments/193/Picture_76.png!

# If your job has been submitted to the cluster, a page will appear with a link "Check status of job", which allows tracking of the job via its log-file.  This link is also accessible from the "1 running" option under the "Run Particle Clustering" submenu in the appion sidebar.

# Once the job is finished, an additional link entitled "1 complete" will appear under the "Run Particle Clustering" tab in the appion sidebar.  Clicking on this link opens a summary of all clustering runs that have been done on this project.    

!http://emg.nysbc.org/attachments/192/Picture_79.png!

h2. Notes, Comments, and Suggestions:

# Clicking on "Show Composite Page" at the top of the Particle Clustering List page will expand the page to show the relationships between alignment runs and feature analysis runs.

______

[[Run Feature Analysis|<Run Feature Analysis]] | [[Run Particle Clustering|Run Particle Clustering >]]

______