annotate kmersvm/train.xml @ 0:7fe1103032f7 draft

Uploaded

<tool id="kmersvm_train" name="Train SVM">
  <description>on regulatory DNA sequences</description>
  <command interpreter="python">scripts/kmersvm_train.py -q -p -s -v $N -C $SVMC -e $EPS 
    #if $weight_type.weight_type_select == "custom"
	  -w $weight_type.weight
	#end if
    #if $kernel.kernel_select == "sk"
      -t 1 -k $kernel.kmerlen_sk
    #else
      -t 2 -k $kernel.kmerlen_wsk -K $kernel.kmerlen_wsk2
    #end if
	$inputA $inputB
  </command>
  <inputs>
    <param format="fasta" name="inputA" type="data" label="Positives"/>
    <param format="fasta" name="inputB" type="data" label="Negatives"/>
    <conditional name="kernel">
        <param name="kernel_select" type="select" label="Kernel Type">
            <option value="sk">Spectrum Kernel</option>
            <option value="wsk">Weighted Spectrum Kernel</option>
        </param>
        <when value="sk">
            <param name="kmerlen_sk" type="integer" value="6" label="K-mer Length">
                <validator type="in_range" message="K-mer length must be in range 5-10" min="5" max="10" />
            </param>
        </when>           
        <when value="wsk">
            <param name="kmerlen_wsk" type="integer" value="6" label="Minimum K-mer Length">
                <validator type="in_range" message="K-mer length must be in range 5-10" min="5" max="10" />
            </param> 
            <param name="kmerlen_wsk2" type="integer" value="8" label="Maximum K-mer Length">
                <validator type="in_range" message="K-mer length must be in range 5-10" min="5" max="10" />
            </param>             
        </when> 
    </conditional>
    <param name="N" type="select" label="N-Fold Cross Validation">
        <option value="3">3</option>
        <option value="5" selected="true">5</option>
        <option value="10">10</option>
    </param>
    <conditional name="weight_type">
    	<param name="weight_type_select" type="select" label="Positive Set Weight">
			<option value="automatic">Automatic</option>
			<option value="custom">Custom</option>
		</param>
		<when value="custom">
    		<param name="weight" type="float" value="1" label="Input The Value of Positive Set Weight" />   
		</when>
    </conditional>
    <param name="SVMC" type="integer" value="1" label="Regularization Param C" />
    <param name="EPS" type="float" value="0.00001" label="Precision Param E" />
  </inputs>
  <outputs>
    <data format="tabular" name="SVM_weights" from_work_dir="kmersvm_output_weights.out" label="${tool.name} on ${on_string} : Weights" />
    <data format="tabular" name="CV_predictions" from_work_dir="kmersvm_output_cvpred.out" label="${tool.name} on ${on_string} : Predictions" />
  </outputs>
   <tests>
        <!--SK-->
        <test>
          <param name="kernel_select" value="sk"/>
          <param name="inputA" value="test_positive.fa" />
          <param name="inputB" value="test_negative.fa" />
          <param name="weight_type_select" value="automatic" />
          <output name="output" file="test_weights.out" compare="re_match" lines_diff="20"/>
          <output name="output2" file="train_predictions.out" compare="re_match"/>	
        </test>
  </tests>
  <help>
  
**Note**
 
.. class:: warningmark
  
All values of K-mer lengths must be between 5 and 10 bp.
  
----
  
**What it does**
  
Takes as input 2 FASTA files, 1 of positive sequences and 1 of negative sequences.  Produces 2 outputs: 
  
  A) Weights: list of sequences of length K ranked by score and posterior probability for that score.
  	
  B) Predictions: results of N-fold cross validation
  
----
  
**Parameters**
  
Kernel: 2 choices:
  
  A) Spectrum Kernel: Analyzes a sequence using strings of length K.
  	
  B) Weighted Spectrum Kernel: Analyzes a sequence using strings of range of lengths K1 - Kn.
  	
N-Fold Cross Validation: Number of partitions of training data used for cross validation.
  
Weight: Increases importance of positive data (increase if positive sets are very trustworthy or for training with very large negative sequence sets).
  
Regularization Parameter: Penalty for misclassification.  Trade-off is overfitting (high parameter) versus high error rate (low parameter).
  
Precision Parameter:  Insensitivity zone.  Affects precision of SVM by altering number of support vectors used.
  
----
  
**Example**
  
Weights file::
  
    #parameters:
    #kernel=1
    #kmerlen=6
    #bias=-1.20239998751
    #A=-1.50821617139
    #B=-0.110516009177
    #NOTE: k-mers with large negative weights are also important. They can be found at the bottom of the list.
    #k-mer	revcomp	SVM-weight
    AGGTCA	TGACCT	9.32110889151
    AAGGTC	GACCTT	8.22598019901
    ACCTTG	CAAGGT	5.78739494153
    AGGTCG	CGACCT	5.40759311635
	
Predictions file::
  
    mm8_chr1_10212203_10212303_+	3.31832111466	1	1
    mm8_chr1_103584748_103584848_+	-0.253869299667	1	3
    mm8_chr1_105299130_105299230_+	-1.03463560077	1	3
    mm8_chr1_106367772_106367872_+	5.36528447025	1	3
  
  </help>
</tool>