Hi-C Data Resolution Improvement Method based on Ensemble Learning

Abstract

The multi-level spatial structure of chromosomes allows remote regulatory elements in the linear coordinate space to closely regulate the expression level of the target genes in the three-dimensional structural space, so, the efficient analysis will be essential. Especially, this paper focuses on the Hi-C data resolution improvement method based on ensemble learning. Hi-C data standardization is used to remove the systematic bias between samples introduced by the various unavoidable nonrandom factors, hence, the accuracy is essential. Therefore, this study utilizes the stacking integration model to achieve the ensemble task, the designed model can avoid the problems of low prediction accuracy and the poor model robustness. Similarly, the multi-objective regression evolved based on the idea of multi-label classification. After testing the designed model on the public data sets, the accuracy can reach more than 99%. Compared with the traditional tools, our designed algorithm reaches better results.