Skip to main content
Fig. 2 | Epigenetics & Chromatin

Fig. 2

From: Integrative prediction of gene expression with chromatin accessibility and conformation data

Fig. 2

The performance of gene expression prediction models measured in terms of Spearman correlation on hold-out test data is shown for various models using peak length, peak count, and peak signal within the gene promoter regions. Two different window sizes (\(3\,\text {kb}\), \(50\,\text {kb}\)) and the nearest gene approach are compared. We observe that the \(50\,\text {kb}\) models outperform the \(3\,\text {kb}\) models. Considering the \(50\,\text {kb}\) models, there is a slight advantage of the window-based models over the nearest gene-based annotation

Back to article page