view cluster.tools/partition.xml @ 0:0decf3fd54bc draft

Uploaded

<tool id="partiton_clust" name="Partition Clustering" force_history_refresh="True">
    <command interpreter="python">partition.py
-d $dataset 
${dist_obj}
-n ${direction} 
-a $alg_cond.algorithm
#if $alg_cond.algorithm == 'pam' # -m ${alg_cond.distance_metric}
#end if
-k ${numk} 
-o ${output}

</command>
    <inputs>
    	<param name="dataset" type="data" format='tabular' label="Data Set"  help="Specify the data matrix (tab-delimited) to be clustered"/>
	<param name="dist_obj" type="boolean" label="Distance Object (R dist object)?" truevalue="-D" falsevalue="" checked="False" help="Check if the matrix contains the pairwise distances between a set of objects"/>

    	<param name="direction" type="select" label="Cluster Columns or Rows?" help="Specify the matrix dimension to cluster (see help below)">
	  <option value="cols">Columns (Samples)</option>
	  <option value="rows" selected='true'>Rows (Genes)</option>
    	</param>
	
	<conditional name='alg_cond'>
	  <param name="algorithm" type="select" label="PAM or K-means?" help="Specify the partition cluster method to use (see help below)">
	    <option value="km">K-means</option>
	    <option value="pam" selected='true'>PAM</option>
	  </param>
	  <when value='pam'>
	    <param name="distance_metric" type="select" label="Distance Metric" help="Specify the distance metric to use (see help below)">
	      <option value="cosine" selected='true'>Cosine</option>
	      <option value="abscosine">Absolute Cosine</option>
	      <option value="pearson">Pearson</option>
	      <option value="abspearson">Absolute Pearson</option>
	      <option value="spearman">Spearman</option>
	      <option value="kendall">Kendall</option>
	      <option value="euclidean">Euclidean</option>
	      <option value="maximum">Maximum</option>
	      <option value="manhattan">Manhattan (AKA city block)</option>
	      <option value="canberra">Canberra</option>
	      <option value="binary">Binary</option>
	    </param>
	  </when>
	</conditional>
    	<param name="numk" type="integer" label="Number of Clusters" value="50" help="Specify the number of clusters to use"/>
    	
    </inputs>
    <outputs>
        <data format="rdata" name="output" label="Partition Clustering Data (RData)"/>
    </outputs>
<help>
.. class:: infomark
     
**Perform Partition Clustering (Cluster Samples) on a specified data set**

----

**Parameters**

- **Data Set** - Specify the data matrix to be clustered.  Data must be formated as follows:

         * Tab-delimited
         * Use row/column headers

- **Distance Object** Specify whether or not the data set is a pairwise distance matrix

- **Cluster Samples or Genes** - Specify the dimension of the matrix to cluster:

         * Rows (Genes)
         * Columns (Samples)

- **PAM or K-means?** Specify which partition clustering method to use - users have choice of:

         * PAM (Partition Around Mediods)
         * K-means

- **Distance Metric** Specify the distance metric to use.  Note, this is ONLY AVAILABLE IF PAM IS THE ALGORITHM BEING USED.  Choice of:

	 * Cosine (AKA uncentered pearson)
	 * Absolute Cosine (AKA uncentered pearson, absolute value)
         * Pearson (pearson correlation)
	 * Absolute Pearson (pearson correlation, absolute value)
         * Spearman (spearman correlation)
	 * Kendall (Kendall's Tau)
         * Euclidean (euclidean distance)
	 * Maximum
	 * Manhattan (AKA city block)
	 * Canberra
	 * Binary


- **Number of Clusters** Specify the number of clusters to use

</help>
</tool>