Gene structure prediction from consensus spliced alignment of multiple ESTs matching the same genomic locus

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Accurate gene structure annotation is crucial but challenging in genomics.
• Existing methods rely on comprehensive cDNA or EST collections, which are often unavailable.
• Previous spliced alignment tools failed with non-cognate ESTs due to high similarity requirements.

Purpose of the Study:

• To develop a computational tool for gene structure annotation using diverse EST resources.
• To overcome the limitations of existing tools in handling non-cognate ESTs from related species.
• To improve gene prediction sensitivity and enable annotation of genomes with incomplete EST data.

Main Methods:

• Development of GeneSeqer, a computer program for aligning thousands of ESTs with long genomic sequences.
• Algorithm designed to tolerate high percentages of mismatches and indels in ESTs.
• Integration of novel splice site prediction models within GeneSeqer.

Main Results:

• GeneSeqer efficiently aligns large numbers of ESTs to genomic sequences.
• The tool successfully uses non-cognate ESTs from homologous genes and related species for gene prediction.
• Demonstrated utility for plant genome annotation, particularly for Arabidopsis thaliana.
• Proposed as a timely tool for rice genome annotation using cross-species ESTs.

Conclusions:

• GeneSeqer provides a robust solution for gene structure annotation, especially when comprehensive EST data is lacking.
• The program's ability to utilize diverse ESTs enhances gene prediction sensitivity.
• GeneSeqer is a valuable tool for annotating plant genomes, including rice, and is available as downloadable code and web servers.