TODO

If score threshold doesn't exist in file, the step 5 breaks

pipeline_type parameter currently doesn't do anything



Check required parameters by step...

convert docker to singularity image
docker run \
>     -v /var/run/docker.sock:/var/run/docker.sock \
>     -v $PWD:/output \
>     --privileged -t --rm \
>     singularityware/docker2singularity \
>     kdrew/msblender

To access local files, wrap them in file() in the process definition. As in the feature_processes.nf
See https://groups.google.com/forum/#!topic/nextflow/KjTp_M4VyKg


FYI  docker build -t clairemcwhite/cfmsflow:latest . 

# Docker done
#Progress on docker... 
#   Start with just getting everything installed
#   Then worry about the external github repositories



# Test that feature selection worked

Where are defaults coming from?

Change help to show default values


#jq -s 'reduce .[] as $item ({}; . * $item)' json_files/*


#Change TPOT to GroupKFold

#Change training output to named objects

# example training still taking > 2 minutes
Add option to limit featsel, and scalingto tpot

Something changed between TPOT 0.9.3 and 0.10.0 to slow it down. Probably scanning more things. Explore ways to speed up

Now requires TPOT >= 0.10.0 

Break up model and applying model steps (break step 4 into two steps)

Check required parameters by step range

Break out gold standard step - OK

Check with python2 and python3

Clean up params

Add option to plot or not

Contrive example ( or do three tries from gold standard set?)

Add final optional annotation
walktrap,and optional elution and annotations