2.6 Identification of TADs and TAD boundaries
The normalized contact matrix was used as input to perform TAD identification as reported in mammalian (Dixon et al., 2012). Directionality index (DI) was calculated from upstream 2 Mb to downstream 2 Mb along the center of each bin at 40 kb resolution and Hidden Markov model (matlab 2013, HMM_calls.m) was then used to predict the states of DI for final TAD generation. We used the same criteria 400 kb (distance between the two adjacent TADs) to distinguish unorganized chromatin with topological boundaries, that was the topological boundaries to be less than 400 kb and unorganized chromatin to be greater than 400 kb. Transcription start sites (TSSs) were calculated based on the number of annotated genes. Firstly, the position of TAD Boundary center was obtained, and then the number of TSS in each 10 Kb window of the upstream and downstream of TAD Boundary Center with the distance delimiting size of 10 Kb was calculated.