This repository contains analysis in Python and R for the output of in vivo perturb-seq experiments as described in Liu et al, bioRxiv 2023.
First, you'll want to install dependencies. Python and R dependencies are managed separately.
Dependencies are managed as an R project through renv
. Activate the project by installing renv
and running:
renv::init()
renv::restore()
in the top-level project directory. This will install dependencies as specified in renv.lock
.
This repository uses Poetry for Python dependency management. Install poetry, navigate to the
project directory, and run poetry install
. This will install all packages according to the versions specified in poetry.lock
.
Next, you'll want to create a top-level output
directory and a top-level data
directory. The code in this repository assumes that
these directories (and sometimes subdirectories) exist. output
generally contains output plots and data, while data
contains
input metadata and raw data. We have chosen to leave most of our existing directory structure following these data
and output
directories intact as examples; there's user flexibility on where things go.
Here is the directory structure as used by us as developers:
.
├── R
│ ├── malignant
│ ├── microenvironment
│ ├── revisions
│ └── utils
├── README.md
├── data
├── genome-biology-methods.Rproj
├── output
├── poetry.lock
├── pyproject.toml
├── python
│ ├── chromatin_analysis.ipynb
│ ├── doubles_coefficient_analysis.ipynb
│ └── gl261_cinemaot.ipynb
├── renv.lock
├── shared_data
│ └── deseq_output
└── shell
R
contains R codepython
contains Python codeshell
contains shell scripts for CUT&TAG analysisshared_data/deseq_output
contains comprehensive DESeq2 outputs for our perturb-seq data relative to non-targeting same treatment condition controls. This data was generated using the RDElegate
package.
Authored by Christopher Zou, Ashir Borah, and John Liu. Feel free to fork/ping us through issues for questions.