Some Miscellaneous Information about the DNAnexus Platform
Job Execution Overview
There are three types of executable on the platform:
- Applets – non optimized programs (not fully vetted by DNAnexus) these are either home-grown or provide on the platform.
- Apps – DNAnexus deployed applications that have been optimized and vetted.
- Workflows – collections of Apps or Applets strung together to perform a complex set of operations.
Job Flow:
Additionally, there are VISUALIZERS that are combinations of html and javascript (do not spawn on a worker node) used for data visualization… these can set up to reference files on the platform.
Data Transfer Methods (All are checked for integrity by MD5SUM check ??)
- Upload/Download via the Web pages
- Upload via the dx-upload tool
- Upload via the incremental upload tool (for “live data” direct from sequencers)
Note: Globus is also supported… details to be forthcoming.
Controls on Job Execution (Safety valves)
- All jobs have a preset max number of nodes (~100).
- Jobs that run more than 24 hours generate warning emails.
- All apps have (should have) reasonable timeout policies.