creativeyann17 / arranger

Data Portal API and component Generation - https://www.overture.bio/products/arranger

Arranger

Generate and manage your own genomic data portal.

Develop (Edge):

Master (Release):

Note: Arranger is undergoing refactoring work, for what will become version 3+. While we do not foresee many dramatic breaking changes, an upgrade guide is already in the works. Current users of v2 may look at the legacy branch meanwhile, where we will continue fixing newly reported bugs until further notice.

You may need to add the --legacy-peer-deps flag when integrating our modules into your apps. Along with the rewrite, we're steadily updating the internal dependencies so this isn't necessary.

Documentation

This file is meant as a quick introduction, but for more in-detail documentation, you should explore Arranger's "Read the Docs". If interested, see our Open Source License

• Getting Started
  • Development Setup
  • Dockerized Setup (a.k.a. Quickstart)
• Motivation
• What is a "Data Portal"?
• Roadmap
• Development Details

---

Getting Started

Arranger has been tested using NodeJS version: ^16.13.2, and it may not work correctly on versions <14

- Development Setup

Setting up the project, and prepare things to make changes

# 1. clone the repository
  git clone [email protected]:overture-stack/arranger.git

# 2. enter the project's folder
  cd arranger

# 3. install the dependencies
  npm ci

# 4. install the module's own dependencies
  npm run bootstrap

Now you should be able to start the following processes from the project's root folder:

# watch all modules and rebuild them when you make changes
  npm run watch

# test all modules at once
  npm run test

# run the server (on port 5050)
  npm run server

# serve the component dashboard (on port 6060)
  npm run dashboard

# serve the component portal (on port 7070)
  npm run portal

# run storybook (on port 8080)
  npm run storybook

- Dockerized Setup

A bit more friendly "quickstart", if you just want to get things started

# Start all services at once, using some default settings.
# This runs the following services: Elasticsearch, Kibana, and Arranger Server:
  make start

# ^^^^ ^^^^^ which stands for the following command behind the scenes:
# ES_USER=elastic ES_PASS=myelasticpassword docker-compose -f docker-compose.yml up -d -build

# Alternatively, you could start the services separately, like so:
  make start-es # Elasticsearch
  make start-server # Arranger Server

# Note: The ES_* environment values may be customized when running your own Arranger instance. They can be found atop the `Makefile`.

---
# Afterwards, in another bash process, you may seed an example/mock file_centric index:
  make seed-es

# ^^^^ ^^^^^^^ which runs the following command behind the scenes:
# ./docker/elasticsearch/load-es-data.sh ./docker/elasticsearch elastic myelasticpassword

# That script file may give you ideas on how to automate uploading indexes to your instance.


---
# Bonus: ----------------------------- #
# See other preprogrammed make targets
  make help
# e.g. utilities to list the indexes, or clear the Elasticsearch; list the running docker containers, etc.

---

Motivation

The Ontario Institute for Cancer Research (OICR) has built a few Data Portals. e.g.:

• International Cancer Genome Consortium (ICGC) Data Portal
• Genomic Data Commons (GDC) Data Portal (joint effort with University of Chicago)

Although they are not identical in architecture, available data or overall purpose, there is tremendous amount of overlap in how they function and how users interact with them, despite being implemented differently. It's no coincidence. The GDC Data Portal was directly influenced by the ICGC Data Portal.

With new projects ahead of us, there is an opportunity to create a framework designed to act as a core library for any given data portal, similar to what Elastic's Kibana accomplishes; but based on the features of our existing portals, and the expectation of continuous improvement and domain specific customization.

There are many potential benefits:

• Reduce duplicate code
• Ability to fix bugs and add features to many projects at once
• Pool developer resources
• Increase cross-team communication
• Encourage open source contribution

---

What Is A "Data Portal"?

Topology

This is a simplified "birds-view" diagram of what goes on under the hood

---

Roadmap

Short Term

• cli tool for bootstrapping new projects

• Provide all necessary modules to implement searching functionality

  • Dynamic GraphQL schema generation
  • API Server (GraphQL endpoint)
  • Query / Aggregation building middleware
  • Response middleware (ie. removing null aggregations)
  • UI Components
    • Aggregations
      • Simple view
      • Advanced View
    • Results Table
    • SQON Display

• Provide editor interface to expose common transformations (similar to the Babel or bodybuilder REPLs)
  • Elasticsearch Mappings -> GraphQL Schema
  • GraphQL Query -> Elasticsearch Queries

Medium Term

• Authentication
• Sets
• Analysis

Long Term

• Kibana Plugin
• Hosted Data Portal generating service

---

Development Details

Arranger is a lerna flavored monorepo. The modules exposed by Arranger compose all of the necessary code required to build an application such as the Genomic Data Commons.*

Releasing Instructions

• From main branch, run npm run tag <version>
• Publishing process will be run by Jenkins

* The GDC contains many features that are out of Arranger's scope