AndersenLab cloud resources

For full documentation visit

Google Domains

Any domain names the lab uses should be registered with Google Domains. The two ones currently are:

Google domains can be used to forward domain-specific email addresses if necessary. For example, could be created and forwarded to an email address.

Google Cloud

Google cloud is used for a variety of services that the lab uses.

Google Cloud Storage

Google cloud storage is used to store and distribute files on CeNDR, for cegwas, and for some files on the lab website.


Files are grouped into 'buckets' on google storage. We use the following buckets:

This bucket contains all the data associated with It is broken down into six primary directories.

  • browser_tracks - for genome-browser tracks that rarely if ever change.
  • db - Storage/access to the SQLite database.
  • photos - sample collection photos.
  • releases - dataset releases. For more detail, see wi-nf.
  • reports - images and data files within reports.
  • static - static assets used by the site.

In some cases the data associated with a publication is too large to put on github. We store those data here, along with a couple other odds and ends.

Other buckets

Google Cloud creates a bunch of other buckets too. Most of these should be ignored as they are part of the

Secret Bucket

There is one other secret bucket. Ask Erik about it.

cegwas (deprecated)

Currently this bucket has a SQLite database used by cegwas. This will be replaced to use the same database used by CeNDR in the db/ folder.

Once the new version of cegwas is developed - this section should be removed, and that bucket should be deleted.

Google datastore

Google datastore used as the database for CeNDR. It stores information on mappings, traits, and more.

App engine

App engine is the platform CeNDR runs on.

Error Reporting

Google cloud contains a really nice error reporting interface. Error reports are generated whenever something goes wrong on CeNDR. A github issue can be created for these errors and they can be addressed.


We have used bigquery in the past for large query jobs. We are not actively using it as of late.



S3 = simple storage service. We use it to store BAMs.


Amazon Fargate is used to run the mapping pipeline on CeNDR