Adapt for AWS -- using the HCP S3 bucket

All the HCP data is available in a public AWS bucket. 

Adapting this code to run on AWS -- from preprocessing to prediction -- seemlessly would greatly enhance its re-usability in the community, and our own further developments. 