view cluster.tools/cluster.2.centroid.xml @ 2:b442996b66ae draft

Uploaded

<tool id="cluster_2_centroid" name="Calculate Matrix of Cluster Centroids (Mean Euclidean)" force_history_refresh="True">
    <command interpreter="python">cluster.2.centroid.py
-d $dataset 
-g ${gen_new_dgram}
-o ${output_fname}

</command>
    <inputs>
    	<param name="dataset" type="data" format='rdata' label="Cluster Result" help="Specify the cluster result to analayze (MUST BE IN rdata format; see help below)"/>
	<param name="gen_new_dgram" type='select' display="radio" label="Re-calculate cluster tree for new matrix ?" help="Specify whether or not to re-calculate a dendrogram for the cluster centroid matrix">
	  <option value="no">No</option>
	  <option value="yes" selected='true' >Yes</option>
	</param>

    </inputs>
    <outputs>
        <data format="rdata" name="output_fname" label="Cluster Centroid Matrix (RData)" />
    </outputs>
<help>
.. class:: infomark
     
**Calculate Matrix of Cluster Centroids** - Tool to calculate a new matrix containing the cluster centroids,  NOTE, this tool will automatically detect whether the dimensionality of the clusters (rows or columns).

**OUTPUT:**  A new CDT file

----

**Parameters**

- **Cluster Result** - Specify the cluster result to analayze (MUST BE IN rdata format), and must contain the same objects that are produced by the 'Partition Clustering,' 'Hierarchical Clustering (HAC),' or 'Consensus Clustering' tools.  Specifically, it must contain the following objects

         * A 'treecl.res' or 'partcl.res' object - corresponding to whether the cluster results is from a partition or tree clustering method
         * A 'data' object that contains the data that was passed into the clustering method.  NOTE, it is better for this to be the actual data passed in, rather than the data prior to the pre-processing that was performed prior to the actual clustering.

- **Re-calculate cluster tree for new matrix?** - Specify whether or not to re-calculate a dendrogram for the cluster centroid matrix.
        * IF the cluster result was a tree cluster, the new dendrogram will use the distance method from the original clustering
	* IF the cluster result was a partition cluster, the dendrogram will use 'Euclidean' distance


</help>
</tool>