view cluster.tools/select.k.from.consensus.cluster.xml @ 9:a3c03541fe6f draft default tip

Uploaded

<tool id="select_consensus_k" name="Select Consensus Clustering K" force_history_refresh="True">
    <command interpreter="python">select.k.from.consensus.cluster.py
-r ${cc_results}
-c ${change_min}
-m $metric_cond.metric
-h $report 
-p ${report.files_path}
#if str($metric_cond.metric) == 'survival':
-s $metric_cond.survival
-e $metric_cond.mode
#end if
-o ${cluster_class}
    </command>
    <inputs>
        <param name="cc_results" type="data" format="rdata" label="Consensus Clustering Results File" help="Specify the result from a consensus clustering run (see help below)"/>
	<conditional name="metric_cond">
	  <param name="metric" type="select" label="Change Metric" help="Specify the choice of 'change' metric to use">
            <option value="difference">Difference</option>       
            <option value="silhouette">silhouette</option>
            <option value="survival">survival</option>
	  </param>
	  <when value="survival">
	    <param name="survival" type="data" label="Clinical Data" optional="true" help="Specify Clinical data to use"/>
	    <param name="mode" type="select" label="Survival Report Mode" help="Mode to use when performing Log-Rank tests">
	      <option value="all">All</option>
	      <option value="one">One</option>
	      <option value="both">Both</option>
	    </param>    
	  </when>
	</conditional>
	<param name="change_min" type="float" label="Minimum relative change (AUC; pivot point for search)" value="0.1" help="Specify threshold to determine the minimum relative change"/>

    </inputs>
    <outputs>
        <data format="rdata" name="cluster_class" label="Select K from Consensus Clustering Report (RData)"/>
        <data format="html" name="report" label="Select K from Consensus Clustering Report (HTML)"/>        
    </outputs>
<help>
.. class:: infomark
     
**Select an choice of K from consensus clustering result**

----

**Parameters**

- **Consensus Clustering Results File** Specify the result from a consensus clustering run (**MUST BE THE RDATA** outuput)

- **Change Metric** Specify the choice of 'change' metric to use.  Choice of:

         * Difference (relative difference in AUC between different choices of K)
         * Silhouette (Max average silhouette width - search anchored by 'Change Min')
         * Survival (Min p-value from Log-rank tests - search anchored by 'Change Min') **NOTE: WE DISCOURAGE USING THIS METRIC.**

- **Change Min** Threshold to determine the minimum relative change that's used to decide the choice of K


- **IF 'SURVIVAL' IS THE CHANGE METRIC, the following become available:**

- **Clinical Data** Specify Clinical data to use for performing the Log-Rank tests if 'Survival' is the metric 
        *  **(MUST SPECIFY A PROPERLY FORMATTED CLINICAL DATA FILE** - See the "Format Raw TCGA sample IDs")

- **Clinical Report Mode** Mode to use when performing Log-Rank tests.  Choice of:

         * All - All clusters versus each other
         * One - One cluster versus a meta-cluster composed of the others.  Search performed exhaustively.
         * Both - Perform both all-v-all and and one-v-others test; select the choice of K that gives the best

</help>
</tool>