ArcFISH as a command line tool#

Here we demonstrate how to use ArcFISH as a command line tool. Make sure arcfish is successfully installed (see installation) and activate the environment with ArcFISH in shell:

>>> conda activate arcfish_env
>>> python -m pip show arcfish
Name: arcfish
Version: ...

Here we use the DNA seqFISH+ data from Takei et al., 2021 to call chromatin loops, TADs, and A/B compartments. The data can be downloaded from the 4DN data portal with ID: 4DNFIW4S8M6J (biological replicate 1), 4DNFI4LI6NNV (biological replicate 2), and 4DNFIDUJQDNO (biological replicate 3).

Download the data and place them like the following:

.
├── data
│   ├── takei_science_2021
│   │   ├── 4DNFIW4S8M6J.csv
│   │   ├── 4DNFI4LI6NNV.csv
│   │   └── 4DNFIDUJQDNO.csv
├── output

Preprocessing#

First run the following to preprocess the data and store them as AnnData objects:

arcfish preprocess \
    -i "rep1::data/takei_science_2021/4DNFIW4S8M6J.csv,\
        rep2::data/takei_science_2021/4DNFI4LI6NNV.csv,\
        rep3::data/takei_science_2021/4DNFIDUJQDNO.csv" \
    -o pp_chr_adata \
    -v X::1000,Y::1000,Z::1000

The -v argument specifies how to convert the input data to nm. Since the 3D coordinates are stored as µm, we multiply each axis by 1000.

This step creates a directory pp_chr_adata with a .h5ad file for each chromosome. The processed data will be used in all following steps.

Loop calling#

Once the processed data are stored in pp_chr_adata, loop calling can be done by

arcfish loop -i pp_chr_adata -o output/loop_res.bedpe

which stores the loop calling result to output/loop_res.bedpe.

If desired, loop calling parameters can be changed by adding additional arguments:

  1. fdr: FDR cutoff to define loop candidates, by default 0.1.

  2. pval: p-value cutoff to filter loop summits, by default 1e-5.

  3. lo: minimum loop length to be considered, by default 100Kb.

  4. up: maximum loop length to be considered, by default 1Mb.

  5. gap: candidates within gap away are treated as from the same summit, by default 50Kb.

  6. outer: local background size, by default 50Kb.

For example, to test loops with range from 10Kb to 2Mb, change the command to

arcfish loop -i pp_chr_adata -o output/loop_res.bedpe -lo 10000 -up 1000000

Domain calling#

To call TADs, run the following

arcfish domain -i pp_chr_adata -o output/domain_res.bedpe

The domain calling result is in output/domain_res.bedpe.

Additional optional arguments for TAD calling are:

  1. fdr: FDR cutoff to define TAD boundaries, by default 0.1.

  2. window: region size to consider intra/inter-domain contacts, by default 100Kb.

  3. tree: whether to return hierarchical TADs, by default True.

  4. min: minimum TAD size, by default 0.

Compartment calling#

To call A/B compartments, run the following

arcfish cpmt -i pp_chr_adata -o output/cpmt_res.bedpe

An optional -min argument can be passed in to specify the minimum A/B compartment size, by default 0.

Note

The compartment calling result here is not very meaningful as only a 1.5Mb (60 loci at 25Kb resolution) region is imaged on each chromosome. This is mainly to demonstrate the compartment calling workflow.