Related Experiment Videos
A hidden Markov model approach for determining expression from genomic tiling micro arrays
Kasper Munch1, Paul P Gardner, Peter Arctander
1Bioinformatics Centre, Institute of Molecular Biology and Physiology, University of Copenhagen, Universitetsparken 15, 2100 Copenhagen, Denmark. Kasper@binf.ku.dk
BMC Bioinformatics
|May 5, 2006
Summary
ExpressHMM, a new probabilistic method, improves the analysis of genomic tiling array data for identifying novel transcribed regions. This adaptive modeling approach enhances accuracy in predicting gene expression and non-coding RNA.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- Genomic tiling microarrays offer potential for discovering novel coding and non-coding RNA.
- Current analysis of tiling array data is often performed in an unstructured, ad hoc manner.
Purpose of the Study:
- To introduce ExpressHMM, a probabilistic procedure for adaptive modeling of genomic tiling array data.
- To improve the prediction of gene expression and transcribed fragments from genomic sequence data.
Main Methods:
- Utilized a hidden Markov model (HMM) to model probe score distributions in expressed and non-expressed genomic regions.
- Trained the HMM on annotated expression and non-expression regions.
- Applied the trained HMM to predict transcribed fragments and generate expression probability curves.
Main Results:
- Successfully applied ExpressHMM to tiling array data from ten human chromosomes.
- Demonstrated the effectiveness of adaptive modeling for fluorescence scores.
- Achieved superior nucleotide sensitivity and transfrag specificity compared to previous methods.
Conclusions:
- Adaptive modeling of fluorescence scores is valuable for categorizing expressed and non-expressed probes.
- ExpressHMM provides a more accurate and sensitive method for analyzing genomic tiling array data.
- The developed approach outperforms existing methods in identifying transcribed regions.