Skip to content

Latest commit

 

History

History
76 lines (58 loc) · 1.75 KB

manual_install.md

File metadata and controls

76 lines (58 loc) · 1.75 KB
Raw

Get the code

First, download and checkout the latest release

# from a directory of your choice
git clone https://github.com/weberlab-hhu/Helixer.git
cd Helixer
# git checkout dev # v0.2.0

System dependencies

Python 3.8 or later

Python development libraries

Ubuntu (& co.)

sudo apt install python3-dev

Fedora (& co.)

sudo dnf install python3-devel

Virtualenv (optional)

We recommend installing all the python packages in a virtual environment: https://docs.python-guide.org/dev/virtualenvs/

For example, create and activate an environment called 'env':

python3 -m venv env
source env/bin/activate

The steps below assume you are working in the same environment.

GPU requirements (optional, but highly recommended for realistically sized datasets)

And to run on a GPU (highly recommended for realistically sized datasets), everything for tensorflow-gpu is required, see: https://www.tensorflow.org/install/gpu

The following has been most recently tested.

python packages:

  • tensorflow-gpu==2.7.0

system packages:

  • cuda-11-2
  • libcudnn8
  • libcudnn8-dev
  • nvidia-driver-495

A GPU with 11GB Memory (e.g. GTX 1080 Ti) can run the largest configurations described below, for smaller GPUs you might have to reduce the network or batch size.

Post processor

https://github.com/TonyBolger/HelixerPost

Setup according to included instructions and further add the compiled helixer_post_bin to your system PATH.

Most python dependencies of Helixer

# from the Helixer directory
pip install -r requirements.txt

Helixer itself

# from the Helixer directory
pip install .  # or `pip install -e .`, if you will be changing the code