EFO Elastic app

This is a practice project integrating Elasticsearch (elasticsearch-browser) with an Angular app.
Example data was extracted from the Experimental Factor Ontology - a hierarchical set of terms describing human diseases and their relations. Each record includes its EFO ID, the official disease name, a list of synonyms/alternative identifiers and the number of associated drug targets.

Technologies:

• Node.js (v.10.16.0)
• Angular (v.8.1.2)
• MDB for Angular (v.8.1.0)
• Elasticsearch (v.7.1.0)
• Elasticsearch Node.js client (v.16.3.0)

Features:

Records can be located by official EFO ID or by disease name. Exact match in disease name is prioritised, followed by partial matches, then matches in known disease synonyms. Results are then weighted by known drug target associations from Open Targets data. The effect of altering the weights given to each can be tested:

Results are returned as the query is entered with the typeahead/autocomplete search:

Learning outcomes:

• Using Rxjs - revision of Promises+Observables
• Elasticsearch - Practise using different search mechanisms.
• Incorporating searches with a user interface, e.g. autocomplete/type-ahead suggestions
  • Keyword and edge ngram analyzers are used for the type-ahead search - a maximum ngram length of 15 was needed, as biomedical terms are approximately 2/3rds longer on average than lay language

Setup:

To run the application using the webpack dev server:

1. Install dependencies (Node.js, Angular CLI etc.)
2. Clone the repository
3. cd efo-elastic-app/
4. npm start
5. Open your browser at http://localhost:4200/

The application expects an Elasticsearch instance running at http://localhost:9200/ with an index titled 'efo'. Use the shell scripts included in the EFO-data folder to create the index and populate it with the processed EFO data.