Skip to content

ncihtan/htan-portal

Repository files navigation

HTAN Data Portal

This repo contains the code for the Human Tumor Atlas Network Data Portal

Framework

This is a Next.js project bootstrapped with create-next-app

Backend

All data is coming from Synapse. We have a Python script that generates a JSON file that contains all the metadata. There is currently no backend, it's a fully static site i.e. all filtering happens on the frontend.

Update Data Files

Update release information

Only certain metadata rows and data files on Synapse are released. We keep track of this information in Google BigQuery. One can get the latest dump of that using these commands (requires access to the htan-dcc google project):

bq extract --destination_format CSV released.entities_v6_0 gs://htan-release-files/entities_v6_0.csv
bq extract --destination_format CSV released.metadata_v6_0 gs://htan-release-files/metadata_v6_0.csv
gsutil cp gs://htan-release-files/entities_v6_0.csv entities_v6_0.csv
gsutil cp gs://htan-release-files/metadata_v6_0.csv metadata_v6_0.csv

Pull files from Synapse and Process for ingestion

cd data
# Run the script that pulls all the HTAN metadata
# It outputs a JSON in public/syn_data.json and a JSON with links to metadata in data/syn_metadata.json
python get_syn_data.py
cd ..
# Find and replace certain values (this is a temp fix)
yarn findAndReplace
# we store the result of this in gzipped format
gzip -c public/syn_data.json > public/syn_data.json.gz
# Convert the resulting  JSON to a more efficient structure for visualization
# Note: we output stdout and stderr to files to share these with others for
# data qc debugging purposes
# TODO: there is a ssl legacy provider is hack for
# https://stackoverflow.com/questions/69692842/error-message-error0308010cdigital-envelope-routinesunsupported
yarn processSynapseJSON > data/processSynapseJSON.log 2> data/processSynapseJSON.error.log
# we also store the processed data in gzipped format
gzip -c public/processed_syn_data.json > public/processed_syn_data.json.gz

Export to bucket

At the moment all data is hosted on S3 for producion. This is because there is a file size limit for vercel. To update it:

  1. gzip file (note that it's already gzipped in the repo)
  2. Remove ".gz" extension so it's just json and rename to include current date in filename.
  3. Upload file to s3 bucket "htanfiles" (part of schultz AWS org)
  4. The file needs two meta settings: Content-Encloding=gzip and Content-Type=application/json
  5. Once file is up, change path in /lib/helpers.ts

Or step 1-4 as command:

MY_AWS_PROFILE=inodb
aws s3 cp processed_syn_data.json.gz s3://htanfiles/processed_syn_data_$(date "+%Y%m%d_%H%M").json --profile=${MY_AWS_PROFILE} --content-encoding gzip --content-type=application/json --acl public-read
aws s3 cp metadata_gzip s3://htanfiles/metadata --recursive --profile=${MY_AWS_PROFILE} --content-encoding gzip --content-type=text/csv --acl public-read

Testing

There are currently no automated tests, other than building the project, so be careful when merging to master

Getting Started

First, make sure you have the latest processed json file:

yarn gunzip

Run the development server:

npm run dev
# or
yarn dev

Open http://localhost:3000 with your browser to see the result.

You can start editing any page. The page auto-updates as you edit the file.

Debugging processSynapseJSON

Add debugger; somewhere in the code. Then run:

node --openssl-legacy-provider ./node_modules/.bin/ncc build --source-map --no-source-map-register  data/processSynapseJSON.ts

Followed by:

node  --inspect-brk dist/index.js

Now you can attach to it in e.g. VSCode

Learn More about Next.js

To learn more about Next.js, take a look at the following resources:

Deployment

The app is deployed using the ZEIT Now Platform from the creators of Next.js.