Overview

SemNet is a series of modules for working with semantic networks. More specifically, we implemented algorithms that would help us gather features and rank nodes in terms of their connections to other nodes. The software was designed to work with a network of biomedical concepts stored in a locally hosted Neo4j database (node types, edge types). Details on all of the algorithms can be found in the original paper.

Installation

To install, clone the repository to your machine, change to the semnet directory, and use pip install .. You should then be able to import semnet and use it.

You can create the graph by installing Neo4j on your computer, downloading the node and edge files (available here) into the Neo4j import directory, and running `neo4j_import.sh`.

Features

The following features have been implemented in SemNet:

• Feature Extraction

  • Metapath counts between node pairs
  • Degree-weighted path counts between node pairs
  • HeteSim similarity scores between node pairs

• Computation

  • Latent semantic relationship extraction
  • User-guided similarity search
  • Unsupervised ranker aggregation

• Utilities

  • Neo4j database interaction
  • Query parallelization
  • Vectorization
  • Negative example generation

Citation

If you use SemNet, please cite us!

@article{sedler_2019_semnet,
  title={SemNet: Using Local Features to Navigate the Biomedical Concept Graph},
  author={Sedler, Andrew R and Mitchell, Cassie S},
  journal={Frontiers in Bioengineering and Biotechnology},
  volume={7},
  pages={156},
  year={2019},
  publisher={Frontiers}
}

References

• Himmelstein, Daniel S., and Sergio E. Baranzini. "Heterogeneous network edge prediction: a data integration approach to prioritize disease-associated genes." PLoS computational biology 11.7 (2015): e1004259.
• Klementiev, Alexandre, Dan Roth, and Kevin Small. "An unsupervised learning algorithm for rank aggregation." European Conference on Machine Learning. Springer, Berlin, Heidelberg, 2007.
• Shi, Chuan, et al. "HeteSim: A General Framework for Relevance Measure in Heterogeneous Networks." IEEE Trans. Knowl. Data Eng. 6.10 (2014): 2479-2492.
• Wang, Chenguang, et al. "Relsim: relation similarity search in schema-rich heterogeneous information networks." Proceedings of the 2016 SIAM International Conference on Data Mining. Society for Industrial and Applied Mathematics, 2016.
• Yu, Xiao, et al. "User guided entity similarity search using meta-path selection in heterogeneous information networks." Proceedings of the 21st ACM international conference on Information and knowledge management. Acm, 2012.