Improved Fourier Transform Method for Unsupervised Cell-Cycle Regulated Gene Prediction
CSB '04 Proceedings of the 2004 IEEE Computational Systems Bioinformatics Conference
Weighted Top Score Pair Method for Gene Selection and Classification
PRIB '08 Proceedings of the Third IAPR International Conference on Pattern Recognition in Bioinformatics
ISBRA'07 Proceedings of the 3rd international conference on Bioinformatics research and applications
International Journal of Data Mining and Bioinformatics
| Hi-index | 0.00 |
Motivation: The abundance of a significant portion ofthe temporal induction-repression expression pattern of agene among other genes in a time-course data is an indicationof its non-randomness. The significance of the portionsthat match between two gene profiles can be derived usingbinomial analysis or its variant. Considering the induction-repressionpattern alone is both meaningful and significantsince the related genes induced/repressed in a given periodmay not show the same exact shape of induction/repression.Further, microarray measurements are of low quality, whichmight make expression patterns of related genes less similar.Based on this observation we developed an algorithm calledfriendly neighbors (FNs). This algorithm finds the significancescore of a gene as the number of genes in the sameexperiment that share its induction-repression pattern morethan a certain threshold. The concept of friendly neighborsis different from that of nearest neighbors. A neighborthat satisfies certain similarity condition is called friendlyneighbor where as a nearest neighbor is the most similarneighbor of all neighbors. This leads to the observation thatall friendly neighbors does not necessarily be nearest neighbors,vice versa.Results: The FNs approach has been applied to discoverputative estrogen target genes and to detect cell cycleregulated genes in S. cerevisiae. The new approach performedbetter than paired t-test and simple expression levelbased filtering methods on estrogen target gene discovery.It did significantly well on cell cycle regulated genediscovery in the absence of task-specific knowledge. Its performanceis better than commonly used fourier transformmethod and fold change methods. Apart from detectingcell cycle regulated genes, it also detected other prominentpatterns which could be detected only by more complicatedclustering and data analysis methods.Availability: http://giscompute.gis.a-star.edu.sg/FNs