Hi!

I want to run a HCL using the Spearman rank correlation and complete linkage.

Doing that manually in the session produces one result, but if I create a script to do the same thing I get a different result with quite a different sample clustering.

It seems that when running a HCL in a script the Euclidean distance metric is always used, no matter what I pick. If I use e.g. an ST in my script instead, I actually get the distance metric I asked for.

Is this a known bug, or am I doing something wrong? Below I have pasted an example of a script that does not produce the expected result.

<?xml version="1.0"?>
<!DOCTYPE TM4ML SYSTEM "C:\Users\christin\.mev\mev_script_dtd.dtd">
<TM4ML version="1.0">
   <mev version="1.0">
      <!- Original Script Creation Date: Apr 21, 2010 2:15:43 PM ->

      <!- Script Name: Test ->

      <!- Script Description: Test ->

      <primary_data id="1"/>
      <analysis>
         <alg_set input_data_ref="1" set_id="1">
            <algorithm alg_id="1" alg_name="HCL" alg_type="cluster" input_data_ref="1">
               <plist>
                  <param key="distance-absolute" value="false"/>
                  <param key="calculate-genes" value="true"/>
                  <param key="calculate-experiments" value="true"/>
                  <param key="method-linkage" value="1"/>
                  <param key="distance-factor" value="1.0"/>
                  <param key="distance-function" value="9"/>
               </plist>
               <output_data output_class="single-output">
                  <data_node data_node_id="2" name="Single Ordered Output"/>
               </output_data>
            </algorithm>
         </alg_set>
      </analysis>
   </mev>
</TM4ML>