Chapter 2 Configure Environment for Full Profiling Pipeline

This workflow assumes you have already set up an AWS account with an S3 bucket and EFS, and created a VM per the instructions in the link below.

2.1 Launch an AWS Virtual Machine for making CSVs and running Distributed-CellProfiler

Launch an EC2 node using AMI cytomining/images/hvm-ssd/cytominer-bionic-trusty-18.04-amd64-server-*, created using cytominer-vm.

You will need to create an AMI for your own infrastructure because the provisioning includes mounting S3 and EFS, which is account specific. We recommend using an m4.xlarge instance, with an 8Gb EBS volume.

Note: Proper configuration is essential to mount the S3 bucket. The following configuration provides an example, named imaging-platform (modifications will be necessary).

  • Launch an ec2 instance on AWS
  • AMI: cytomining/images/hvm-ssd/cytominer-ubuntu-trusty-18.04-amd64-server-1529668435
  • Instance Type: m4.xlarge
  • Network: vpc-35149752
  • Subnet: Default (imaging platform terraform)
  • IAM role: s3-imaging-platform-role
  • No Tags
  • Select Existing Security Group: SSH_HTTP
  • Review and Launch
  • ssh -i <USER>.pem ubuntu@<Public DNS IPv4>

After starting the instance, ensure that the S3 bucket is mounted on ~/bucket. If not, run sudo mount -a.

Log in to the EC2 instance.

Enter your AWS credentials

aws configure

The infrastructure is configured with one S3 bucket. Mount this S3 bucket (if it is not automatically mounted)

sudo mount -a

Check that the bucket was was mounted. This path should exist:

ls ~/bucket/projects

2.2 Create a tmux session

You will want to retain environment variables once defined, and for processes to run when you are not connected, so you should create a tmux session to work in

tmux new -s sessionname

You can detach from this session at any time by typing Ctl+b, then d.
To reattach to an existing session, type tmux a -t sessionname

You can list existing sessions with tmux list-sessions and kill any poorly-behaving session with tmux kill-session -t sessionname

2.3 Define Environment Variables

These variables will be used throughout the project to tag instances, logs etc so that you know which machines are working on what, which files to operate on, where your logs are, etc.




MAXPROCS=3 # m4.xlarge has 4 cores; this should be # of cores on your instance - 1

2.4 Create Directories

mkdir -p ~/efs/${PROJECT_NAME}/workspace/

cd ~/efs/${PROJECT_NAME}/workspace/

mkdir -p log/${BATCH_ID}

2.5 Download Software

cd ~/efs/${PROJECT_NAME}/workspace/
mkdir software
cd software
git clone
git clone

cd ..

If these repos have already been cloned, git pull to make sure they are up to date.

This is the resulting structure of software on EFS (one level below workspace):

└── software
    ├── Distributed-CellProfiler
    └── pe2loaddata