nf-co.re

This repository contains code for the nf-core website: http://nf-co.re/

Packages used

Here's how the website is built:

• Language: PHP
• HTML / CSS / JS framework: Bootstrap v5
• JavaScript libraries:
  • jQuery
  • Popper.js (used for bootstrap tooltips)
  • highlightjs (syntax highlighting)
  • Leaflet (contributor map)
  • Moment.js (time and date parsing)
  • Chart.js (statistics plots)
  • hammer.js (mobile touch interaction handling)
  • chartjs-plugin-zoom (Zoom and pan plugin for Chart.js)
  • Canvas2Svg.js (SVG exports of Chart.JS plots)
  • FileSaver.js (Trigger browser downloads from in-page data, used to save plot SVGs to files)
  • jQuery table sorter (sorting tables)
• PHP Markdown parsing: Parsedown and Parsedown Extra
• SVG icons: http://www.flaticon.com, https://worldvectorlogo.com/

Development

Getting the code

To make edits to the website, fork the repository to your own user on GitHub and then clone to your local system.

git clone git@github.com:[USERNAME]/nf-co.re.git
cd nf-co.re/

Running a local server

Ok, you're ready! To run the website locally, just start the apache-php server with:

docker compose up

You should then be able to access the website in your browser at http://localhost:8888/.

If you prefer, you can also use a tool such as MAMP - if so, set the base directory to /path/to/nf-co.re/public_html in Preferences > Web-Server > Document Root and then hit Start Servers.

Most of the hand-written text is in /markdown, to make it easier to write. The PHP files in /public_html then parse this into HTML dynamically, if supplied with a filename.

Note that the .htaccess file is set up to remove the .php file extensions in URLs.

First-run

Much of the site is powered by a pipelines.json file. The webserver does this automatically when GitHub events trigger an update, but you'll need to run the script manually.

Access tokens

First you'll need a config.ini text file with values for github_username and github_access_token. See instructions on how to get a GitHub OAuth token (the token only needs the public_repo permission). This file is ignored in .gitignore for security reasons.

For the MySQL database you should also add the following values:

host = 'db'
port = '3306';
dbname = 'nfcore';
username = 'nfcore_admin';
password = 'PEBBLY8exhibit_mead1cilium6despise'

Running PHP scripts

It's easiest to run these first manual update scripts on the command line. If you have PHP available then you may be able to do this directly. Alternatively, if you are using Docker as above then you can open a shell inside the running container. The container is typically named web (you can check this with the docker ps command), so you can open an interactive shell using the following command:

docker exec -it web /bin/bash
cd var/www/

Update scripts

The following command will create public_html/pipelines.json, which is used by the website.

php update_pipeline_details.php

To update the modules database (from within the docker container) run:

docker exec -it nf-core-web /usr/local/bin/php /var/www/update_module_details.php

Note that this is also ignored in the .gitignore file and will not be tracked in git history.

Optionally, once you've done that, you can grab the pipeline traffic, issue statistics and font awesome icons:

php update_issue_stats.php
php update_stats.php
php update_fontawesome_icons.php

Note that your GitHub account needs push rights for the nf-core permission for the update_stats.php to work.

This creates nfcore_stats.json, nfcore_issue_stats.json and public_html/assets/js/fa-icons.json, all also ignored in .gitignore.

Production Server Setup

Deployment

The website is deployed via GitHub Actions (.github/workflows/web-deploy.yml). This script runs PHP composer and npm, then syncs the required files to the web server via FTP.

Tools API docs

Tools docs are built using GitHub Actions on the nf-core/tools repo using Sphinx. These actions sync the built HTML files via FTP.

GitHub web hooks

There is a GitHub web hook at the nf-core organisation level which triggers the pipeline update script whenever a repo is created, or has a release etc. This pings the deploy_pipelines.php script.

Stats cronjob

The web server needs the following cronjobs running to scrape statistics and udates:

0 0 * * * /usr/local/bin/php /path/to/deployment/update_stats.php >> /home/nfcore/update.log 2>&1
0 2 * * * /usr/local/bin/php /path/to/deployment/update_issue_stats.php >> /home/nfcore/update.log 2>&1
0 0 * * 0 /usr/local/bin/php /path/to/deployment/update_fontawesome_icons.php >> /home/nfcore/update.log 2>&

Remember to replace /path/to/deployment/ with your actual deployment directory.

The update_issue_stats.php script can use a lot of GitHub API calls, so should run at least one hour after the update_stats.php script last finished. This is not because the script takes an hour to run, but because the GitHub API rate-limiting counts the number of calls within an hour.

Contribution guidelines

If you are looking forward to contribute to the website or add your institution to the official list of contributors, please have a look at the CONTRIBUTING.md.

Community

If you have any questions or issues please send us a message on Slack.

Credits

Phil Ewels (@ewels) built the website, but there have been many contributors to the content and documentation. More recently, @mashehu has done a great deal of work with the code. See the repo contributors for more.

Kudos to the excellent npm website, which provided inspiration for the design of the pipeline pages.