https://github.com/vam-sin/ribogl
Graph+Sequence approach to predicting ribosome densities at the gene level
https://github.com/vam-sin/ribogl
graph-neural-networks interpretability lstm ribosome-profiling
Last synced: 6 months ago
JSON representation
Graph+Sequence approach to predicting ribosome densities at the gene level
- Host: GitHub
- URL: https://github.com/vam-sin/ribogl
- Owner: vam-sin
- Created: 2024-04-04T16:21:00.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2024-04-04T17:01:14.000Z (about 1 year ago)
- Last Synced: 2024-04-05T17:55:25.364Z (about 1 year ago)
- Topics: graph-neural-networks, interpretability, lstm, ribosome-profiling
- Language: Python
- Homepage:
- Size: 17.6 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-riboseq - Code
README
# RiboGL
Repository of "Towards improving full-length ribosome density prediction by bridging sequence and graph-based representations". The dataset files can be found here: https://zenodo.org/records/12180308
## RiboGL Model
## Plotly Server
### Setup
Inside the `interpret` subfolder one can find the code to run the Plotly server to intepret the RiboGL predictions.
In order to run the server:
1. Install the dependecies in a conda envirorment `ribogl` (arbitrary name):
```bash
conda env create -n ribogl -f interpret/requirements.yaml
```
2. Download the data from https://zenodo.org/records/12180308 into the `data` subfolder.
3. Extract `RiboGL_Attributions.h5.gz` into `RiboGL_Attributions.h5` using `gzip -d RiboGL_Attributions.h5.gz`. The uncompressed size will be around 43 GB.
4. Run the server:
```bash
conda activate ribogl
python interpret/app.py --port 8050
```
The server will be available at http://localhost:8050/. The port can be changed with the `port` argument.