Related Experiment Videos
Speeding disease gene discovery by sequence based candidate prioritization
Euan A Adie1, Richard R Adams, Kathryn L Evans
1Medical Genetics Section, Department of Medical Sciences, The University of Edinburgh, Edinburgh, UK. euan.adie@ed.ac.uk
BMC Bioinformatics
|March 16, 2005
Summary
Machine learning identifies disease-associated genes using sequence features. The PROSPECTR tool prioritizes candidate genes, improving efficiency in genetic studies.
Area of Science:
- Genetics
- Bioinformatics
- Machine Learning
Background:
- Genetic linkage studies often identify large regions containing numerous genes.
- Traditional methods rely on functional annotation, which can be limiting.
- Disease-associated genes exhibit distinct sequence-based features.
Purpose of the Study:
- To develop an automated method for prioritizing candidate genes within large genomic regions.
- To leverage sequence-based features for improved gene prioritization in disease association studies.
Main Methods:
- Analysis of sequence-based features differentiating disease genes from non-disease genes.
- Development of the PROSPECTR classifier using the alternating decision tree algorithm.
- Ranking genes based on their likelihood of involvement in human hereditary diseases.
Main Results:
- Significant differences were observed in sequence features between disease-associated and non-disease genes.
- PROSPECTR classifier was created based on these differentiating features.
- PROSPECTR achieved a two-fold enrichment of disease genes 77% of the time, a five-fold enrichment 37% of the time, and a twenty-fold enrichment 11% of the time.
Conclusions:
- PROSPECTR offers a simple, effective method for identifying genes in Mendelian and oligogenic disorders.
- The tool outperforms existing sequence-based classifiers on new data.
- PROSPECTR can significantly reduce the time and effort for researchers investigating large genomic regions.