Molclass

Moleculair classification tool, enter your search query, and results from multiple sources (dbSNP, Franklin, Clinvar, etc) will be aggregated

Molclass is available as a Docker container:
docker run -p 8080:8080 ghcr.io/erasmusmc-bioinformatics/molclass:latest

And connect to Molclass on http://127.0.0.1:8080.

HMF

The Hartwig Medical Foundation has a curated list of variants that can be used as a source, to enable it set the path to the database file with the HMF_DATABASE environment variable, for example with Docker:

docker run \
    -p 8080:8080 \
    -v /host/path/to/hmf_hotspots.tsv:/databases/hmf_hotspots.tsv \
    -e HMF_DATABASE=/databases/hmf_hotspots.tsv \
    ghcr.io/erasmusmc-bioinformatics/molclass:latest

OncoKB

OncoKB can be used as a source if you have an API key, it can be found under your account settings.
The OncoKB API key can be added through the ONCOKB_API_KEY environment variable, for example with Docker:

docker run \
    -p 8080:8080 \
    -e ONCOKB_API_KEY=<api-key> \
    ghcr.io/erasmusmc-bioinformatics/molclass:latest

Alamut

Molclass can also connect to the Alamut API, this requires a computer running the Alamut client that Molclass can query.
For Molclass to connect to this Alamut client, 3 environment variable need to be set:

• ALAMUT_IP: the IP address of the machine running the Alamut client
• ALAMUT_INSTITUTION: Your institution ID
• ALAMUT_API_KEY: Your API key

For example with Docker:

docker run \
    -p 8080:8080 \
    -e ALAMUT_IP=127.0.0.1 \
    -e ALAMUT_INSTITUTION=<institution-id> \
    -e ALAMUT_API_KEY=<api-key> \
    ghcr.io/erasmusmc-bioinformatics/molclass:latest

Cosmic

To use Cosmic, the Mutant Census database must be downloaded and configured. The database can be downloaded [https://cancer.sanger.ac.uk/cosmic/download] if you have an account, from the "Census Genes Mutations" section, then set the COSMIC_DATABASE environment variable to the path of the downloaded file, for example with Docker:

docker run \
    -p 8080:8080 \
    -v /host/path/to/Cosmic_MutantCensus_v98_GRCh37.tsv.gz:/databases/Cosmic_MutantCensus_v98_GRCh37.tsv.gz \
    -e COSMIC_DATABASE=/databases/Cosmic_MutantCensus_v98_GRCh37.tsv.gz \
    ghcr.io/erasmusmc-bioinformatics/molclass:latest

Production

On every new release a Github action will trigger a new Docker build for ghcr.io/erasmusmc-bioinformatics/molclass.
The production environment uses this image in docker-compose.yml to run the Molclass production server.
The production Molclass server is running at CIT, first ssh to the jumpoff server, creig.erasmusmc.nl, then ssh to p-molclass01. Execute a docker-compose pull to pull the newly released image and then docker-compose up -d --build to restart the service with the updated release image.

SSL certificate update

Make a copy of the old cert/key:

cp /etc/nginx/certs/bifrost.erasmusmc.nl.crt /etc/nginx/certs/bifrost.erasmusmc.nl.crt.2023.old
cp /etc/nginx/certs/bifrost.erasmusmc.nl.key /etc/nginx/certs/bifrost.erasmusmc.nl.key.2023.old

Overwrite with the new cert/key:

mv bifrost.erasmusmc.nl.crt /etc/nginx/certs/bifrost.erasmusmc.nl.crt
mv bifrost.erasmusmc.nl.key /etc/nginx/certs/bifrost.erasmusmc.nl.key

Restart the httpproxy docker:

docker restart httpproxy

Dev

Structure

Molclass is a FastAPI webserver where, after an initial page load, the client connects back to the webserver to a websocket to allow for asynchronous querying/updating of sources.

Sources define what meta data they need to be able to complete a request and Molclass will iteratively call sources as more meta data becomes available.

Each source adds the meta data it gets to the pool of available meta data.
It is possible that sources return different meta data. For example, searching NM_000267.3:c.5286T>A where Clingen and Franklin both prefer NM_001042492.3:c.5349T>A.
The meta data in the search always has precedence over the sources, but in other cases Molclass counts how many sources agree and picks the majority.

Simple search example

If only the dbSNP and Clingen sources were active and the user searched for rs1057519915, Molclass would:

In the first iteration only query dbSNP because Molclass only has an rs#.
It adds the transcript (NM_004958.4) and cdot (c.7500T>G) to the meta data pool (+other meta data)

In the second iteration Molclass can now call Clingen with the transcript/cdot values it got from the first iteration.

The third iteration Molclass will see that there are no more sources left, and end the iteration.

Uvicorn start

uvicorn main:app --reload --port 8585 --host 0.0.0.0

Typescript compile

tsc static/molclass.ts --outfile static/molclass.js --watch --lib 'es2020,dom'

Build PyInstaller

Activate the venv:
env/Scripts/Activate.ps1

And run: pyinstaller --clean molclass.spec