Related Experiment Videos
Support vector machines for separation of mixed plant-pathogen EST collections based on codon usage
Caroline C Friedel1, Katharina H V Jahn, Selina Sommer
1Institut fuer Informatik, Ludwig-Maximilians-Universitaet Muenchen, Oettingenstrasse 67, 80538 Muenchen, Germany. friedel@informatik.uni-muenchen.de
Bioinformatics (Oxford, England)
|December 9, 2004
Summary
This study introduces Eclat, a novel computational method for accurately identifying the origin of plant and fungal gene sequences. Eclat utilizes codon usage bias and machine learning to achieve high classification accuracy, improving upon existing methods for expressed sequence tag analysis.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Identifying host and pathogen genes at the plant-pathogen interface is crucial for understanding plant-microbe interactions.
- Challenges in sequence identification include short sequence lengths, limited public data, and ambiguous homology between plant and pathogen genes.
Purpose of the Study:
- To develop a novel, reliable, and high-throughput method for classifying the genome origin of expressed sequence tags (ESTs).
- To overcome limitations of existing methods that rely on homologous genes or public database availability.
Main Methods:
- Development of a novel method based on codon usage bias differences between plant and fungal genes.
- Application of Support Vector Machines (SVMs) for sequence origin classification.
- Comparison of SVMs with other machine learning techniques and a probabilistic algorithm (PF-IND).
Main Results:
- The developed software, Eclat, achieved a classification accuracy of 93.1% for Hordeum vulgare and Blumeria graminis ESTs.
- Eclat significantly outperformed PF-IND, which had an accuracy of 81.2% on the same dataset.
- Eclat can reliably classify EST sequences with at least 50 nt of coding sequence and can be trained for various host-pathogen combinations.
Conclusions:
- Eclat provides a robust and accurate method for classifying EST origin, independent of homologous gene availability.
- The software offers a significant advancement in high-throughput analysis of plant-pathogen interactions.
- Eclat is freely available, promoting wider accessibility and application in biological research.