https://github.com/baranzinilab/graphrain
https://github.com/baranzinilab/graphrain
Last synced: 12 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/baranzinilab/graphrain
- Owner: BaranziniLab
- License: apache-2.0
- Created: 2023-01-24T04:26:04.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2023-02-28T05:21:11.000Z (about 3 years ago)
- Last Synced: 2025-05-07T20:44:08.035Z (12 months ago)
- Language: Python
- Size: 465 KB
- Stars: 5
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# graphRAIN
This repository holds the script for graphRAIN algorithm - algorithm that creates explainable node embeddings of a graph.
## Introduction
GraphRAIN stands for Graph Relational Attribute Integrated Node-embedding
This generates embedding vectors for nodes in a graph by incorporating the attributes hanging from the relationship between the nodes. Since the algorithm normalizes the relational attributes, this works for both heterogeneous and homogeneous graphs.
## How to use
Following snippet shows how to use this package:
```
from embedding import RAIN
NCORES =
NBATCH =
metadata_dict = dict(
edge_path = /path/to/edges.tsv,
edge_metadata_path = /path/to/edge_metadata.tsv,
node_metadata_path = /path/to/node_metadata.tsv,
embedding_save_path = /path/to/save/output/files,
graph_path = embedding_save_path/name_of_graph.joblib,
)
embedding_nodeId_list = List of unique nodeIds of graph whose embedding vectors need to be computed
rain = RAIN()
rain.batch_embedding(
metadata_dict = metadata_dict,
embedding_nodeId_list = embedding_nodeId_list,
ncores = NCORES,
nbatch = NBATCH
)
```
**Refer to [run_rain.py](https://github.com/BaranziniLab/graphRAIN/blob/main/run_rain.py) for a full example which you can run in your machine.**
## Sample data provided
[Sample data](https://github.com/BaranziniLab/graphRAIN/tree/main/sample_data) provided is a network of disease and symptom nodes. This is a subnetwork taken from a bigger network called [SPOKE](https://spoke.rbvi.ucsf.edu/).
The objective of providing this sample data is to let user know about the format of the input files that one should provide to run RAIN algorithm.