Chrom-Lasso:A lasso regression-based model to detect functional chromatin interactions using Hi-C data

We present Chrom-Lasso, an algorithm to identify chromatin interactions from Hi-C data, which performs better in identifying functional chromatin interactions that regulates gene expression when comparing with other methods. To futher assess its performance in investigating biological questions and its ability to deal with noisy data, we use an in vitro mouse CD8+ T cell activation model to generate Hi-C data and RNA-seq data at 4 status(Tn, Teff1, Teff2, Tex) during the process of activation, and we sort T cells at different status by gating criteria as follows: Tn(CD8+, CD44-, CD62L+, at day0), Teff1(CD8+, PD-1+, TIM-3-, at day2), Teff2(CD8+, PD-1+, TIM-3-, at day5), and Tex(CD8+, PD-1+, TIM-3+, at day5), for each status, we prepare 1 Hi-C library following the in situ Hi-C prorocol, and we also prepare 3 RNA-seq libraries from 3 times of parallel experiments. Since functional interactions influence the downstream gene expression level, so the relevance between interactions detected by Chrom-Lasso and gene expression would be powerful evidence proving the capability of Chrom-Lasso in detecting functional interactions. Overall design: In situ Hi-C was used to probe the interactions on 3D genome in 4 different status during the process of in vitro mouse CD8+ T cell activation, and RNA-seq was used to profile the change of gene expression during this process.