Go to https://aws.amazon.com/hpc and click on the link that says Create a Free Account, then either sign in with your existing Amazon account, or create a new one. You will then be asked to enter your credit card information, and to pass a security verification. Go ahead and do that now. Finally, you will be asked to choose a Support Plan. Choose the Basic plan, which is free.
You will next be greeted with a message welcoming you to Amazon Web Services. Click on Launch Management Console to get to work.
Go to the console page, click on your instance, and a bunch of information about it will appear at the bottom of the page. Find the part that says Public DNS and copy the address. It will look something like this: ec2-54-244-25-91.us-west-2.compute.amazonaws.com.
Now we will follow instructions very similar to those provided for launching a jupyter-notebook on an Institutional HPC cluster, but with slight variations for the EC2 cluster. One important difference is that because the EC2 instance is temporary, you cannot store your data for long-term on its cloud platform. Therefore we need to transfer the data to the instance when we start, assemble it, and then transfer all of the results back to our local machine when finished, otherwise they will be deleted (or you can pay to keep them stored online, but I'll assume you are cheap like me.)
### Choose an instance type. The free tier plan only let's you connect to one CPU with up to 30GB of storage, and XXGB of RAM. There are many different paid options, however, that offer much greater storage and compute power. In general, you will want at least 1GB of RAM per CPU. So let's say we want 32 CPUs, >60GB RAM, and >1TB of storage. We can look on the website http://www.ec2instances.info to find which instance offers the best price for these requirements. In this case it looks like it is probably the Cluster Compute Eight Extra Large Instance, which costs $1.090 per hour. If you had a data set with 96 individuals of RAD data sequenced on two lanes of Illumina HiSeq I expect you could finish your assembly in <24 hours using this setup. Therefore, your RAD-seq assembly would cost you less that $24. Not too bad.
### Get setup. Download the aws command-line tools using pip
pip install awscli
Then follow the setup instructions here to create credentials for your login: http://docs.aws.amazon.com/cli/latest/userguide/cli-chap-getting-set-up.html#cli-signup
### CLI remote commands...
### local commands...
### Step 1: login to the EC2 instance.