Algorithm 1: 4DN feature analyzer
Input: Hi-C matrices ${\mathbf{A}}^{(m)}\in {\mathbb{R}}^{n\times n}$, and RNA-seq vectors ${\mathbf{r}}^{(m)}\in {\mathbb{R}}^{n\times 1}$, $m=1,\dots ,T$
Output: Low dimensional space ${\mathbf{Y}}^{(m)}$ and genes in loci with the largest structure-function changes
1 Compute degree, eigenvector, betweenness, and closeness centrality of ${\mathbf{A}}^{(m)}$, and define as ${\mathbf{b}}_{deg}^{(m)}$, ${\mathbf{b}}_{eig}^{(m)}$, ${\mathbf{b}}_{bet}^{(m)}$, ${\mathbf{b}}_{close}^{(m)}$, respectively, where each ${\mathbf{b}}^{(m)}\in {\mathbb{R}}^{n\times 1}$
2 Compute the first principal component (PC1) of ${\mathbf{A}}^{(m)}$
3 Form the feature matrices ${\mathbf{X}}^{(m)}=[{\mathbf{b}}_{deg}^{(m)},{\mathbf{b}}_{eig}^{(m)},{\mathbf{b}}_{bet}^{(m)},{\mathbf{b}}_{close}^{(m)},{\mathbf{r}}^{(m)}]$, where ${\mathbf{X}}^{(m)}\in {\mathbb{R}}^{n\times 5}$
4 Normalize the columns of ${\mathbf{X}}^{(m)}$
5 Compute the common low dimensional space ${\mathbf{Y}}^{(m)}$
6 Visualize the low dimensional projection ${\mathbf{Y}}^{(m)}$ or 4DN phase plane
Return: ${\mathbf{Y}}^{(m)}$ and genes in loci with the largest structure-function changes