A comparison of topologically associating domain callers based on Hi-C data

This Github repository is used for reproducing the results in the manuscript A comparison of topologically associating domain callers based on Hi-C data. If you have any questions, please feel free to contact with me (kun.liu AT mail.csu.edu.cn).

Docker images for 26 TAD callers used in this study are deposited at https://hub.docker.com/repository/docker/kmiles18/tad-callers-most-methods and https://hub.docker.com/repository/docker/kmiles18/tad-callers-based-on-matlab. In the docker images, only the test scripts for each TAD caller are available.

Prepare Hi-C data (scripts are in the prepare_data folder)

Prepare the Hi-C data for Section 3.1

step 1. Download merged sequencing reads from Hi-C experiments

./download_raw_data.sh

step 2. Down-sample reads from the pooled Hi-C data and generate KR normalized HI-C matrices using Juicertools

./downsample.sh

Prepare the Hi-C data for Section 3.2 and 3.3

step 1. Download raw hic files for the human GM12878, IMR90, and K562 cell lines

./download_raw_hic.sh

step 2. Generate KR normalized HI-C matrices using Juicertools

./dump_data_from_hic.sh

The synthetic Hi-C data used in this manuscript is deposited at work/simulate_data folder and work/benchmarks.

Call TADs (scripts are in the work folder)

The preprocessed data for Section 3.1 could be deposited at work/GM12878_downsample_diff_reso folder.

The preprocessed data for Section 3.2 and 3.3 could be deposited at work/Rao folder. Directories structure in work/Rao folder

---Rao/

------HIC001/

------HIC002/

------...

Bash scripts in the folder of each TAD caller (such as work/Armatus) shall call TADs from the Hi-C data. The executive file path or command statement should be updated according to your actual situation. It will spend a lot of time for some methods to call TADs, such as TADtree.

Reproduce the results in manuscript

step 1. Decompress the zip files in work folder

step 2. Convert the TADs identified by various TAD callers to TADs with a uniform format

./extract_all.sh
./extract_diff.sh

For convenience, the processed TADs with uniform format would be deposited at work/all_TADs/loci and work/all_TADs/bin folders, if the all_TADs.zip is decompressed.

step 3. Generate figures and tables in the manuscript

Scripts for Figure1,2,3,4,6, Supplementary Figure1,2,3,4,11, Table 2,3 4, and Supplementary Table 3,4,11-17 in work/Fig_Tab folder can directly plot corresponding figures and tables.

Run_TADadjRsquared.sh, Supp_Fig4.py, and Table5_Supp_Table11_12_13_14_15_16.py in work/Fig_Tab folder can generate corresponding figure and tables.

run_rep_size_contacts.sh and Fig6.py in work/Fig_Tab folder can generate Figure6.

run_enrichment.sh and Table2_3_4_Supp_3_4.py in work/Fig_Tab folder can generate corresponding tables.