Computational Analysis of Gene Identification with SAGE

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Serial Analysis of Gene Expression (SAGE) offers genome-wide gene expression profiling without prior transcript knowledge.
• Gene identification in SAGE is challenging due to multiple sequence matches in reference databases like UniGene Human.

Purpose of the Study:

• To evaluate gene identification accuracy using SAGE with the UniGene Human database.
• To assess the impact of tag length and database multiplicity on SAGE gene identification.
• To explore methods for improving SAGE gene identification specificity.

Main Methods:

• Analysis of tag distributions for varying tag lengths from UniGene Human.
• Tag-to-sequence mapping evaluation using a SAGE tag set from human myeloid cells.
• Assessment of the Gene Length Gene Identification (GLGI) method for extending SAGE tags.

Main Results:

• The extensive dbEST component of UniGene leads to significant multiplicity issues, hindering gene identification.
• Longer SAGE tags (hundreds of bases) are necessary for reliable gene identification with UniGene Human.
• The GLGI method, extending SAGE tags, effectively reduces multiple sequence matches.
• A correlation was observed between multiple match severity and high copy number in the myeloid sample.

Conclusions:

• UniGene Human's current structure presents challenges for accurate SAGE-based gene identification.
• Extending SAGE tags using methods like GLGI is crucial for improving sequence specificity.
• Findings provide insights into optimizing UniGene Human as a reference for SAGE gene discovery.