Skip to content

Latest commit

 

History

History
81 lines (62 loc) · 4.34 KB

README.md

File metadata and controls

81 lines (62 loc) · 4.34 KB
Raw

HiChew

What is it?

HiChew is a tool to find optimal TADs or TADs boundaries segmentation and perform time-series clustering on them.

Why HiChew is useful?

You do not need to adjust gamma parameter in segmentation methods Armatus and Modularity, or window parameter in TAD boundaries calling insulation score method. HiChew do this job – it finds gamma or window parameter by adjusting them to the expected size of TADs in your data. The expected TAD size parameter is the main parameter you pass to HiChew.

HiChew also makes time-series clustering of TAD segmentation or TAD boundaries annotation.

Installation

Directly

git clone https://github.com/encent/hichew
cd hichew
pip install -e .

Additionally install Lavaburst package:

git clone https://github.com/nvictus/lavaburst
cd lavaburst
make build -f Makefile
make install -f Makefile

Using Docker

git clone https://github.com/encent/hichew
cd hichew/docker && make build -f Makefile

And then run one of the containers:

docker run -it --rm  -p 9990:9990 -v $(pwd):/hichew --name hichew-bash hichew-bash

or (jupyter):

docker run -it --rm  -p 9999:9999 -v $(pwd):/hichew --name hichew-jupyter hichew-jupyter

How to use it?

Jupyter

See examples directory.

Command line

Command line scripts are located in the cli directory (see examples below). You may use standard setup (see Installation) Just launch docker container and then one of the provided scripts run_segmentation.py and run_clustering.py inside the container (scripts have parameters -- please go inside the code, there are some insights and documentation).

Example for command-line usage (modularity or armatus):

cd hichew/cli
python3 run_segmentation.py -it coolfiles -ip ../data/coolfiles/E-MTAB-4918.sdrf -e DEMO_MODULARITY_60kb -eps 1e-1 -s 3-4h_repl_merged_5kb -res 5000 -chr X,2L,2R,3L,3R -m modularity -g 0,200.0,0.1 -e_mts 60000 -mis 2 -mts 1000 -pcnt 99.9 -vbc 1000
python3 run_clustering.py -sp ../data/experiments/DEMO_MODULARITY_60kb/opt_tads_modularity_60kb_5kb.csv -it coolfiles -ip ../data/coolfiles/E-MTAB-4918.sdrf -e DEMO_MODULARITY_60kb -mode range -m kmeans -nc 15 -s nuclear_cycle_12_repl_merged_5kb,nuclear_cycle_13_repl_merged_5kb,nuclear_cycle_14_repl_merged_5kb,3-4h_repl_merged_5kb -chr X,2L,2R,3L,3R -pcnt 99.9 -rs 42 -res 5000
python3 run_clustering.py -sp ../data/experiments/DEMO_MODULARITY_60kb/opt_tads_modularity_60kb_5kb.csv -it coolfiles -ip ../data/coolfiles/E-MTAB-4918.sdrf -e DEMO_MODULARITY_60kb -mode certain -m kmeans -nc 7 -s nuclear_cycle_12_repl_merged_5kb,nuclear_cycle_13_repl_merged_5kb,nuclear_cycle_14_repl_merged_5kb,3-4h_repl_merged_5kb -chr X,2L,2R,3L,3R -pcnt 99.9 -vbc 1000 -rs 42 -vs 3-4h_repl_merged_5kb -res 5000
# see all flags description and other documentation in code!

Example for command-line usage (insulation):

cd hichew/cli
python3 run_segmentation.py -it coolfiles -ip ../data/coolfiles/E-MTAB-4918.sdrf -e DEMO_INSULATION_60kb -eps 0.05 -s 3-4h_repl_merged_5kb -res 5000 -chr X,2L,2R,3L,3R -m insulation -g 0,200,1 -e_mts 60000 -mis 3 -mts 1000 -pcnt 99.9 -vbc 100
python3 run_clustering.py -sp ../data/experiments/DEMO_INSULATION_60kb/opt_tads_insulation_60kb_5kb.csv -it coolfiles -ip ../data/coolfiles/E-MTAB-4918.sdrf -e DEMO_INSULATION_60kb -mode range -m kmeans -nc 15 -s nuclear_cycle_12_repl_merged_5kb,nuclear_cycle_13_repl_merged_5kb,nuclear_cycle_14_repl_merged_5kb,3-4h_repl_merged_5kb -chr X,2L,2R,3L,3R -pcnt 99.9 -rs 42 -ins True -res 5000
python3 run_clustering.py -sp ../data/experiments/DEMO_INSULATION_60kb/opt_tads_insulation_60kb_5kb.csv -it coolfiles -ip ../data/coolfiles/E-MTAB-4918.sdrf -e DEMO_INSULATION_60kb -mode certain -m kmeans -nc 7 -s nuclear_cycle_12_repl_merged_5kb,nuclear_cycle_13_repl_merged_5kb,nuclear_cycle_14_repl_merged_5kb,3-4h_repl_merged_5kb -chr X,2L,2R,3L,3R -pcnt 99.9 -vbc 100 -rs 42 -vs 3-4h_repl_merged_5kb -ins True -res 5000
# see all flags description and other documentation in code!

Citing

Bykov N.S., Sigalova O.M., Gelfand M.S., Galitsyna A.A. (2020) HiChew: a Tool for TAD Clustering in Embryogenesis. In: Cai Z., Mandoiu I., Narasimhan G., Skums P., Guo X. (eds) Bioinformatics Research and Applications. ISBRA 2020. Lecture Notes in Computer Science, vol 12304. Springer, Cham. https://doi.org/10.1007/978-3-030-57821-3_37