Using Database Matches with HMMGene for Automated Gene Detection in Drosophila

AUTOR(ES)

Krogh, Anders

FONTE

Cold Spring Harbor Laboratory Press

RESUMO

The application of the gene finder HMMGene to the Adh region of the Drosophila melanogaster is described, and the prediction results are analyzed. HMMGene is based on a probabilistic model called a hidden Markov model, and the probabilistic framework facilitates the inclusion of database matches of varying degrees of certainty. It is shown that database matches clearly improve the performance of the gene finder. For instance, the sensitivity for coding exons predicted with both ends correct grows from 62% to 70% on a high-quality test set, when matches to proteins, cDNAs, repeats, and transposons are included. The specificity drops more than the sensitivity increases when ESTs are used. This is due to the high noise level in EST matches, and it is discussed in more detail why this is and how it might be improved.

Documentos Relacionados

• Smart Health Surveillance with Automated Database Using Android Mobile Device
• Erratum: Smart Health Surveillance with Automated Database Using Android Mobile Device
• Automated assembly of protein blocks for database searching.
• Drosophila Genomic Sequence Annotation Using the BLOCKS+ Database
• Detection kinetics for positive blood culture bottles by using the VITAL automated system.