Related Experiment Videos
Identification of human gene functional regions based on oligonucleotide composition
1Department of Cell Biology, Baylor College of Medicine, Houston, TX 77030, USA.
Proceedings. International Conference on Intelligent Systems for Molecular Biology
|January 1, 1993
Summary
This study develops a new computational method to accurately identify coding and intron regions in human DNA sequences. The approach uses oligonucleotide preferences to predict splice sites with high accuracy, improving gene structure analysis.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- Identifying coding and intron regions in genomic DNA is a significant challenge.
- Existing methods for gene structure prediction require improvement in accuracy and efficiency.
Purpose of the Study:
- To develop a computational function for accurately locating coding regions in anonymous human DNA sequences.
- To create a joint splice site prediction scheme that surpasses current methods.
Main Methods:
- Analysis of a large database of human coding and noncoding sequences from GenBank.
- Utilizing oligonucleotide preferences (e.g., 8-mer, 9-mer) within sliding windows (54 bp, 108 bp) to differentiate sequence types.
- Developing a joint splice site prediction scheme incorporating octanucleotide preferences and triplet frequencies near splice junctions.
Main Results:
- Achieved 87-91% accuracy in separating coding/noncoding regions using 9 bp oligonucleotides.
- Reached 89-95% accuracy in separating coding/intron regions using 8 bp oligonucleotides.
- The joint splice site prediction scheme demonstrated 96-97% accuracy, outperforming a complex artificial neural network approach.
Conclusions:
- Oligonucleotide composition provides a robust basis for distinguishing coding, noncoding, and intron regions.
- The developed joint splice site prediction scheme offers a highly accurate and efficient tool for gene structure prediction.
- The findings suggest a model for splicing involving poly-G(C) rich exon flanking sequences and highlight compositional differences in gene regions.