HuBMAP & SPARC Linkages

A tool to discover and visualize SPARC and HuBMAP ontology and data linkages

Product demo

The latest build of this product is at https://sparc-fair-codeathon.github.io/hubmap-link/.

2-minute demo

20210824_2min_presentation.mp4

Presentation (PDF)

Project Goals

This application enables visualization and reporting on the linkages between HuBMAP and SPARC. It performs the following functions:

1. Visualization of the anatomical structures ontologically tagged by SPARC and highlight their linkage with HuBMAP CCF / ASCT+B ontology terms
2. Illustrate how those common terms map to the 3D HuBMAP VHCCF reference system
3. Map samples from SPARC to HuBMAP
4. Visualize the term overlaps
5. Visualize the samples both from HuBMAP and SPARC in the same 3D reference system (CCF).

This study of the interlinkages between HuBMAP and SPARC ontologies and data could be very valuable in longitudinal studies and to further the goals of both HuBMAP and SPARC in cross-consortium interoperability.

See the Project Description, Workflow Document, and Project Presentation for more background information and initial resources for development.

Tracking Development

Development is tracked on the Sprint Backlog. A CHANGELOG details the new developments with each release.

Continuous Integration / Continuous Deployment (CI/CD)

This application is built automatically using GitHub Actions (status). Each push to the main branch starts a workflow that builds the latest version of the code and deploys the built application to GitHub Pages (here). In addition, the API server is updated with each push to the main branch from a configured Heroku instance (here).

Installation

This application consists of two parts, a webapp and an API server. The webapp is dependent on the API server for data transformation and to obscure the SciCrunch API key.

To get started locally, follow these instructions:

1. If you haven't done it already, make a fork of this repo.
2. Clone to your local computer using git.
3. Make sure that you have Node 14.0 installed. See instructions here.
4. Install your IDE of choice. We recommend Visual Studio Code.
5. Install the Angular CLI
   npm install -g @angular/cli

Quick Start

To quickly get a local installation running, this requires running two terminals.

In the first, run:

cd api
npm ci
npm run build-ts

# Uncomment if on a linux or mac-based OS:  
# export SCICRUNCH_API_KEY='APIKEYXXXX'

# Uncomment if on a Windows-based OS:  
# set SCICRUNCH_API_KEY=APIKEYXXXX

npm start

The local API server will be running at http://localhost:5000.

In the second, run:

cd website
npm ci
npm start

The local webapp will be running at http://localhost:4200 and will connect to the local API server.

Building, Development, and Deployment

See the api and website directories for detailed information on how to install, build, develop, and deploy the projects.

Project structure

• /website/ - Parent directory of the hubmap-link web application
• /website/e2e/ - Configuration directory for application end-to-end testing
• /website/src/ - Project source code and assets
• /website/src/app/ - Angular application source code
• /website/src/app/components/ - The web component hierarchy
• /website/src/app/interfaces/ - Typescript interfaces
• /website/src/app/services/ - Modules to maintain inter-component state, data, and data-wrangling business logic
• /website/src/assets/ - Static web assets such as images and datafiles
• /website/src/environments/ - Build environment configuration
• /api/ - The API server to obscure api key and perform some datatype conversions
• /api/src/ - API server source code and assets
• /api/public/ - Publically exposed API server resources
• /api/public/scaffolds/ - SPARC Scaffolds converted to CCF-compatible formats
• /scaffold-to-csv/ - Documentation and raw files demonstrating how to convert SPARC Scaffolds to CCF-compatible formats
• /docs/ - Documentation resources

Data sources

Data Type	Source	Method
HuBMAP ontology concepts	CCF ASCT+B Reporter ASCT+B API	The API references production versions of the HuBMAP CCF ASCT+B tables where concepts are curated
SPARC AS ontology concepts	FDI Lab SciCrunch/SciGraph API	Organ anatomical structure concept datasets are extracted via neighbor search referencing inbound PART_OF predicate up to 10 degrees
SPARC CT ontology concepts	FDI Lab SciCrunch/SciGraph API	Cell Type concepts are extracted via cypher query on UBERON-to-Cell Ontology single-hop linkages across all predicates
HuBMAP dataset metadata	HuBMAP Elastic Search API	Dataset metadata is ingested directly from the endpoint
SPARC dataset metadata	SPARC Elastic Search API	Dataset metadata is filtered to human-species subset

Ontology concept merging

HuBMAP and SPARC concepts are merged principally based on shared ontology IDs. In the case of SPARC, anatomic structures are uniformly annotated with UBERON-sourced IDs. Most HuBMAP concepts are likewise annotated with UBERON-sourced IDs. However, a small proportion of HuBMAP concepts lack an ontology ID assignment and require additional disambiguation. Where HuBMAP concepts lacked an ontological identifier, the concept name is compared to all exact synonyms among participating SPARC nodes. If no matching synonyms exist in SPARC, then the concept is treated as novel and added to the graph.

SPARC Scaffold to HuBMAP CCF Reference Organ Conversion

For detailed instructions on how to convert a scaffold to be CCF-compatible, check out the scaffold-to-ccf directory.

Links and acknowledgements

Team Members:

• Bruce W. Herr II (@bherr2)
• Samuel O'Blenes (@soblenes32)