Abstract
Understanding how genes are functionally related requires efficient algorithms to model networks from expression data. We report a heuristic search algorithm called Two-Level Simulated Annealing (TLSA) that is more likely to find me global optimal network structure compared to conventional simulated annealing and other searching schemes. We have applied this method to search for a global optimised network structure from a synthetic data set and an expression data set of S. cerevisiae mutants. We have achieved better precision and recall compared to other searching algorithms and are able to map relationships more accurately among functionally-linked genes.
Original language | English (US) |
---|---|
Pages (from-to) | 170-186 |
Number of pages | 17 |
Journal | International Journal of Bioinformatics Research and Applications |
Volume | 3 |
Issue number | 2 |
DOIs | |
State | Published - 2007 |
Keywords
- ANOVA
- Bioinformatics
- Correlation
- DNA microarrays
- Differential expression analysis
- Replication
- Sources of variation
ASJC Scopus subject areas
- Biomedical Engineering
- Health Informatics
- Clinical Biochemistry
- Health Information Management