- Article -
| In Silico Biology 4, 0041 (2004); ©2004, Bioinformation Systems e.V. |
Prediction of IFNγ regulated gene transcription
Hong Liu1+*, Nanxiang Ge2+, Kin-Tak Yu1, David Krolikowski1, Asher Zilberstein1 and Chang S. Hahn1
1 Immunology Platform,
2 Department of Biostatistics, Aventis Pharmaceuticals Inc., Route 202/206,
Bridgewater, NJ 08807 USA
+ These two authors contributed equally to this work
* Corresponding author; Email: hong.liu@aventis.com
Edited by E. Wingender; received July 16, 2004; revised and accepted August 26, 2004; published September 10, 2004
Abstract
IFNγ, a cytokine promoting cell-mediated immunity and antiviral effects, regulates the expression of a large set of genes involved in the immune response. Based on logistic regression, an in silico model for predicting IFNγ regulated transcription has been developed by scoring the transcription factor binding sites on the putative promoters of regulated versus not regulated genes derived from the microarray data of IFNγ treated human macrophages. The model effectively discriminates the transcription factor binding sites that confer responsiveness to IFNγ from those that do not. The model has 65% true positive and 22% false positive rates when evaluated on a small validation set. In order to identify potential IFNγ regulated genes in the whole genome, the model has been used to screen 13,668 promoter pairs of human-mouse orthologs/homologs from Ensembl, and 1,387 of them were predicted to be potentially regulated by IFNγ. In the pilot experiment, the regulation pattern of a subset of predicted genes that were not detected by microarray approach was evaluated by quantitative PCR. The results for the four novel genes, which are up regulated by IFNγ in human macrophages and identified by this approach, are described in the present communication.
Key words: IFNγ, transcription, transcription factor binding site (TFBS), promoter, logistic regression model