CoLLAGE

Codon Likelihoods Learned Against Genome Evolution (CoLLAGE): a deep learning framework for identifying naturally selected patterns of codon preference within a species.

Using CoLLAGE

Web

The easiest way to use CoLLAGE is through our free web service located at WEBSITE_URL_TBD. You simply need to upload your FASTA, and the website handles the rest!

Running locally

If you would like to run CoLLAGE locally, please install the dependencies in either cpu_requirements.txt or cuda_requirements.txt depending on whether CUDA is available on your system.

TODO(auberon): Add instructions on running scripts for CoLLAGE.

Running the Docker image

TODO(auberon): give full instructions Input file must be named input.fasta. TODO(auberon): remove this limitation.

docker run -v /absolute/path/to/input/folder:/input -v /absolute/path/to/output/folder:/output redcliffesalaman/collage-model

Hint: to get an absolute path of a folder in your current working directory, you can use:

$(pwd)/my_relative_folder

Developing

If you would like to contribute to the development of CoLLAGE, please see CONTRIBUTING.md for information on how to install and set up your dev environment.