Meta-reg: A Computational Metaheuristic Framework to Improve SVM-Based Prediction of Regulatory Activity

摘要

Gene regulatory activity prediction is an important step to understand which Transcription Factors (TFs) are important for regulation of gene expression in cells. The development of recent high throughput technologies and machine learning approaches allow us to archive this task more efficiently. Support Vector Machine (SVM) has been successfully applied for the case of predicting gene regulatory activity in Drosophila embryonic development. Here, we introduce meta-heuristic approaches to select the best parameters for regulatory prediction from transcription factor binding profiles. Experimental results show that our approach outperforms existing methods and the potentials for further analysis beyond the prediction.