Reranking candidate gene models with cross-species comparison for improved gene prediction
2008

Improving Gene Prediction with Cross-Species Comparison

Sample size: 7777 publication Evidence: high

Author Information

Author(s): Liu Qian, Crammer Koby, Pereira Fernando CN, Roos David S

Primary Institution: University of Pennsylvania

Hypothesis

Can cross-species comparisons improve the accuracy of gene prediction models?

Conclusion

Reranking gene models with cross-species comparison improves gene prediction accuracy.

Supporting Evidence

  • Reranking based on cross-species comparison outperformed the best gene models identified by Evigan alone.
  • ReRanker-5g selected the highest probability Evigan-5g model for 6031 loci.
  • ReRanker-5g improved performance for genes with more exons compared to Evigan-5g.

Takeaway

This study shows that looking at similar genes in other species can help scientists find genes more accurately in a target species.

Methodology

The study used a gene finder called Evigan to generate candidate gene models and then reranked these models based on comparisons with orthologous genes from closely related species.

Potential Biases

Potential bias may arise from the reliance on existing gene models and the training data used for the reranking algorithm.

Limitations

The reranking strategy assumes that a locus contains a single gene, which may not always be the case.

Digital Object Identifier (DOI)

10.1186/1471-2105-9-433

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication