Filtering Alignments

Filtering reads alignments

WARNING: This documentation is outdated and will soon be updated.

Introduction

Eoulsan define a ReadAlignmentsFilter interface that allow to filter reads alignments. To handle read alignments use the picard library.

Using ReadAlignmentsFilter

The ReadAlignmentsFilter interface define a filterReadAlignments method witch takes as unique argument a list with all the alignments for a read (a list of SAMRecord objects). This method remove from the list the alignments that no pass the test. In the next sample, we assume that all the alignments for a read are together in the SAM file.

// Create the reader
final SAMFileReader inputSam =
        new SAMFileReader(new FileInputStream("alignments.sam"));

// Define the filter
ReadAlignmentsFilter filter = new KeepOneMatchReadAlignmentsFilter(); 

String lastId = null;

// A list with all the alignments of a read
List<SAMRecord> records = new ArrayList<SAMRecord>();

for (SAMRecord samRecord : inputSam) {

  String id = samRecord.getReadName();
  if (id==null)
    continue;


  if (!id.equals(lastId)) {

     // Filter the alignments for a same read
     filter.filterReadAlignments(records);

     // Print the filtered alignments
     for (SAMRecord filteredSamRecord: records) {
        System.out.println(filteredSamRecord);
     }

     // Clear the list of alignments     
     records.clear();
     lastId = id;
  }

  alignments.add(samRecord);
}
reader.close();

Example of filterReadAlignments() method

The following filterReadAlignments() method keep only one alignment by read (the first).

  @Override
  public void filterReadAlignments(final List<SAMRecord> records) {

    if (records == null || records.size() < 2)
      return;

    final SAMRecord first = records.get(0);
    records.clear();
    records.add(first);
  }

Writing a plug-in for the filtersam step of Eoulsan

WARNING: filtersam step in Eoulsan 1.1 is not modular. The registering of a ReadAlignmentsFilter will only have an effect in the next version of Eoulsan.

It is very easy to write a plug-in for the filtersam step of Eoulsan. In this section we will write a MyQualityReadAlignmentsFilter class that filter on mean quality.

• First add getName() and getDescription() method to your new filter:

package com.example;

public class MyUniqueReadAlignmentsFilter extends AbstractReadAlignmentsFilter {

 @Override
  public String getName() {

    return "myuniquealignment";
  }

  @Override
  public String getDescription() {
    return "My unique alignment ReadAlignmentFilter";
  }

}

• Then add setParameter method that allow to configure our filter:

@Override
  public void setParameter(final String key, final String value)
      throws EoulsanException {

    // This filter do not take parameter
    // See the ReadFilter documention for more information about this method
  }

• And an init() method to initialize the plug-in once all the parameters has been set. Here for our example, if no threshold has been set throw an exception.

  @Override
  public void init() {

    // Do nothing
    // See the ReadFilter documention for more information about this method
  }

• Now we can add the accept() method:

  @Override
  public void filterReadAlignments(final List<SAMRecord> records) {

    if (records == null || records.size() < 2)
      return;
    
    records.clear();
  }

• Now our ReadAlignmentsFilter can compile and can be used in a standalone program but not as a filtersam plug-in. To enable our ReadAlignmentsFilter as a plug-in we must register it by adding the full name of the class in the fr.ens.transcriptome.eoulsan.bio.readsfilters.ReadAlignmentsFilter text file in the META-INF/services directory. See the Writing Step Plugin for more information:

com.example.MyUniqueReadAlignmentsFilter