swabseq-analysis

Turns kkovary/swabseq_aws into a containerized Flask API with authentication.

Original code has been edited in the following ways:

• deleted the option to push results to github
• instead, it just writes results to local disk, in provided directory path (which will be used by Flask API wrapper to write to a random temp directory)
• added lab-grid/script-runner
• added a Dockerfile that logs into Basespace to be able to get files, and installs the necessary R and Python dependencies.

Development

To run the server locally:

docker-compose up --build

To test, run 2 scripts. One to generate results for the demo sequencing data, and the other to retrieve those results (use .ps only if using Microsoft Powershell):

./test_unauthenticated.sh
<record the id returned>
<wait several minutes until server stops printing processing messages>
./test_unauthenticated-results.sh <id> > demo_output.json

Before running first time, create a .env file:

cp example.env .env

Before running first time, if you will pull sequencing data from Basespace, generate a default.cfg file:

docker-compose run --rm server bs auth \
    --scopes "BROWSE GLOBAL,READ GLOBAL,CREATE GLOBAL,MOVETOTRASH GLOBAL,START APPLICATIONS,MANAGE APPLICATIONS" \
    --force

This will create a default.cfg file in the ./.basespace directory. Future calls to docker-compose up will use the credentials saved in the ./.basespace directory.

Original Script Usage instructions for demo script:

• Rscript countAmpliconsAWS.R --basespaceID [ID for run] --threads [number of threads for running bcl2fastq]
• The --basespaceID is used to identify the run on BaseSpace and then download the raw data which is then demultiplexed with bcl2fastq and then analyzed, where a PDF of run info and results is generated, along with a csv file with the unique DNA barcodes for each sample, the location of that sample on 96 and 384 well plates, the number of counts for the targeted amplicons, and the classification of the sample (COVID positive, COVID negative, or inconclusive/failed sample).