Using native and syntetically mapped cDNA alignments to improve de novo gene finding

Using native and syntetically mapped cDNA alignments to improve de novo gene finding

January 7, 2008 | Mario Stanke*, Mark Diekhans, Robert Baertsch and David Haussler
The article presents an improved gene-finding method, AUGUSTUS, which incorporates multiple sources of evidence to enhance the accuracy of gene predictions. The authors use expressed sequence tags (ESTs) and syntenically mapped alignments from related genomes to improve the quality of gene predictions. They demonstrate that using only ESTs, AUGUSTUS can correctly predict at least one splice form in 57% of human genes, and with additional evidence from other species and human mRNAs, this accuracy increases to 77%. The method is particularly effective for genomes closely related to those with well-annotated genomes or extensive transcript evidence. The article also discusses the integration of alternative splicing information and the handling of ambiguous EST alignments. Overall, the enhanced AUGUSTUS system outperforms existing methods in terms of accuracy, especially when limited resources are available for gene prediction.The article presents an improved gene-finding method, AUGUSTUS, which incorporates multiple sources of evidence to enhance the accuracy of gene predictions. The authors use expressed sequence tags (ESTs) and syntenically mapped alignments from related genomes to improve the quality of gene predictions. They demonstrate that using only ESTs, AUGUSTUS can correctly predict at least one splice form in 57% of human genes, and with additional evidence from other species and human mRNAs, this accuracy increases to 77%. The method is particularly effective for genomes closely related to those with well-annotated genomes or extensive transcript evidence. The article also discusses the integration of alternative splicing information and the handling of ambiguous EST alignments. Overall, the enhanced AUGUSTUS system outperforms existing methods in terms of accuracy, especially when limited resources are available for gene prediction.
Reach us at info@study.space