SEQUAL: a feature based quality evaluator for DNA sequencing

SEQUAL is a spesific way of arranging a couple of algorithms and scripts in such a way that a quality label is generated for a given DNA sequencing trace file. For more informaion please visit http://timewarping.org/sequal or consult the README files.

• README
• data
  • data/README
  • data/labels.txt
• src
  • src/README
  • src/createfeatures.m
  • src/knowledgeflow.kfml
  • src/features.arff
  • src/weka.jar
  • src/gridSearch.jar

Quick Start

• Download src/knowledgeflow.kfml, src/weka.jar, src/gridSearch.jar, and src/features.arff to a same folder.
• Make sure you have the latest Java Runtime Environment.
• On the command line, run the following command:
  • java -classpath weka.jar:gridSearch.jar weka.gui.GUIChooser (on Linux,Mac)
  • java -classpath weka.jar;gridSearch.jar weka.gui.GUIChooser (on Windows)
• Click KnowledgeFlow
• Open (Ctrl-O) knowledgeflow.kfml (fourth icon from right)
• Click Run button
• Right click the TextViewer box to open the results.

Dataset

We used a subset of DNA trace files freely available for noncommercial use from InSNP. We converted them to Standard Chromatogram Format (SCF) by using Phred. We didn't include the trace files because of legal issues. However The manually created quality labels of the SCF files can be found in data/labels.txt.

Features

The SCF files are transformed into a set of features by using the MATLAB script src/createfeatures.m The features as well as the quality labels of the SCF file are saved into src/features.arff.

Classification with WEKA

In order to classify the data with different machine learning algorithms, KnowledgeFlow Environment of WEKA is used. The KnowledgeFlow script is saved to src/knowledgeflow.kfml