HiSTra 1.3

Spectral translocation detection of HiC matrices.

HiSTra

Installation

Dependency

conda create -n HiSTra python=3.8 
conda activate HiSTra
conda install numpy scipy pandas=1.3.5 matplotlib seaborn h5py
conda install -c conda-forge -c bioconda cooler=0.8.11
pip install matplotlib-venn 

Linux OS

pip install HiSTra

Preparation

Download juicer_tool and deDoc. Because of the update of the two softwares, we recommend that you download them from this repo. You can find relevant jar files in the HiSTra/juice and HiSTra/deDoc, respectively.

Make sure chromosome.sizes file is exactly the file used in generating test(and control) sample(.hic or .mcool). Or the error will occer at the early step. Make sure that no underscores('_') is included in the chromosome name.

Directory tree

A recommended work directory looks like:

mkdir work_dir
cd work_dir
mkdir hic_input
# Then move corresponding hic file here.
mkdir TL_output
ln -s deDoc_dir_path .
ln -s juice_dir_path .

Finally, the directory tree is:

├── deDoc
│   ├── deDoc.jar
├── hic_input
│   ├── Control_GSE63525_IMR90_combined_30.hic
│   ├── Test_GSE63525_K562_combined_30.hic
│   ├── Control_GSE63525_IMR90_combined_30.mcool
│   └── Test_GSE63525_K562_combined_30.mcool
├── juice
│   ├── juicer_tools_2.09.00.jar
└── TL_output

Example

Samples

You can download test case from GSE63525. The test sample hicfile is K562 and control sample hicfile is IMR90.

And you can choose the test sample and control sample by yourself.

Resolution

For samples of human, the hicfile should contain 100k and 500k resolution matrix data. In general, the appropriate resolution could be calculated as following:

res_{unit} = 10^{len(max(chromosome_{size}))-4}.

For example, in the hg.sizes the largest size of chromosome is chr1(249250621), the suggested resolution unit would be 100k, and the lower one is defined as:

5 \times res_{unit}.

Command

# Assume you are in the work_dir,a standard command is for hic format file
HiST -t hic_input/Test_GSE63525_K562_combined_30.hic -c hic_input/Control_GSE63525_IMR90_combined_30.hic -o TL_output/ -d deDoc/deDoc.jar -j juice/juicer_tools_2.09.00.jar -s sizes/chrom_hg19.sizes
# or mcool format file
HiST -t hic_input/Test_GSE63525_K562_combined_30.mcool -c hic_input/Control_GSE63525_IMR90_combined_30.mcool -o TL_output/ -d deDoc/deDoc.jar -s sizes/chrom_hg19.sizes
# or mixed format file
HiST -t hic_input/Test_GSE63525_K562_combined_30.mcool -c hic_input/Control_GSE63525_IMR90_combined_30.hic -o TL_output/ -d deDoc/deDoc.jar -j juice/juicer_tools_2.09.00.jar -s sizes/chrom_hg19.sizes
# Then you can find the result in folder TL_output/SV_result.

Figure

An example of TL result with

FAQ

If you meet "Resource temporarily unavailable" or "error: too many open files" or "ValueError: cannot convert float NaN to integer" or "EmptyDataError: No columns to parse from file"?

If your workstation is configured with more than 128 GB of memory and the number of threads is more than 48, you can try the following operations:

1. You can try command ulimit -u 381152 . Here 381152 could be replaced by any big number.
2. You only need to run HiST command a few more times.

These errors usually occur when the input data is mcool. We will fix them in the next version.

If you meet "No such file or directory"...

1. Check matrix_from_hic directory, if no files in sub-directory, check the juicer_log to find error or check cooler package.
2. If juicer_log suggest like "invalid chromosome chr12", you should check the sizes file, a common problem is to tell "chr1" or "1".