Removing and Suppressing Data

What Is The Difference Between Cancellation And Suppression?

Which one applies depends on whether your records have ever been public.

ENA’s policy is that data released into the public domain should remain public. Once a record has been released it becomes part of the permanent scientific record, is mirrored across the INSDC partners, and is picked up by downstream services ENA does not control. Removal after release is therefore an exception rather than a routine operation, and every request is assessed by a curator before it is carried out. Records which have never been public are different: these can be cancelled on request, and doing so is a normal part of correcting a submission.

Cancellation applies to records which are still private and have never been released. Cancelled records are never made public. They remain visible in your Webin account with a status of ‘Cancelled’ so that your submission history stays complete.

Suppression applies to records which are already public. Suppressed records are removed from search where possible, but they remain retrievable by accession number and cannot be recalled from third parties who obtained them while they were public.

For the circumstances in which each availability status applies, see the Data Availability Policy.

If you are not sure which applies to your records, check their status in the Webin Portal before contacting the ENA helpdesk.

What Should I Include In My Request?

Send your request to the ENA helpdesk and include:

• the accession of every record you want removed;

• what should happen to records associated with them, and in particular anything you want kept;

• an explanation of why removal is needed;

• a new release date, if you are asking for temporary suppression rather than permanent removal.

Be explicit about scope. A request to remove ‘the study’ is ambiguous: it may mean the study record alone, or the study together with its samples, runs and analyses. If you ask for part of a study to be removed, ENA will normally check whether you meant only the accessions you listed or everything associated with the study, and this exchange delays the request.

How Do I Remove Records That Have Never Been Public?

Private records can be cancelled. This covers records which were registered by mistake, are duplicates, were created for testing, or are simply no longer needed.

Contact the ENA helpdesk with the accessions and a clear statement of what should be cancelled. Cancelled records will not be released on their release date.

Private objects other than assemblies can also be cancelled programmatically without contacting the ENA helpdesk, using the CANCEL action in the submission XML. See Submission XML: cancel objects.

Note that CANCEL propagates: cancelling a study also cancels its associated experiments and analyses, and cancelling an experiment also cancels its associated runs. If you only want part of a study removed, target the individual objects rather than the study. This route cannot be used on public records.

How Do I Remove Records That Are Already Public?

Public records can only be suppressed by ENA, and only where the request meets the criteria in the Data Availability Policy. Circumstances in which suppression is considered include incorrect or contaminated sequence, incorrect source metadata, a misidentified source organism, a sample mix-up, an erroneous submission, and data released earlier than the submitter intended.

Contact the ENA helpdesk with the accessions and an explanation. A curator will assess the request and carry it out if it is deemed suitable.

Suppression may be permanent or temporary. If you need the records to be released again later, say so and provide the new release date; the data will then be re-released automatically when that date is reached. Contact the ENA helpdesk in good time if that date subsequently needs to change.

What Happens To Data Once It Is Suppressed?

Suppressed records are removed where possible from direct search tools, including text search, sequence similarity search and the ENA APIs. They remain available by accession number, and the ENA Browser will still display the record, marked as suppressed, to anyone who looks it up by its exact accession. Suppressed metadata are decoupled from the other components of the study.

Search indexes are not updated immediately; allow a day or two for a suppression to be reflected in the browser and in search results.

Suppression does not undo release. Data which have been public for even a short time may already have been downloaded, mirrored to the other INSDC partners, or incorporated into downstream databases. ENA cannot guarantee that this has not happened, and cannot control how such copies are subsequently used.

Can I Remove Only Part Of My Submission?

Yes — removal is per record, so you can remove some objects and keep others. What survives depends on the object type.

Reads, Runs And Experiments

Runs and experiments can be removed without affecting the samples they reference. This is the usual route when the wrong files were submitted, when files were assigned to the wrong run, or when a submission failed validation and needs to be replaced.

The sample accessions remain intact and can be reused for the corrected submission. When you resubmit, use experiment and run aliases which do not clash with the removed submission; the file names themselves can be reused. The corrected submission will receive new run and experiment accessions.

If you also need the samples removed, ask for that explicitly and list their accessions.

Samples

Sample records can be cancelled where they were registered by mistake, are duplicates, or are no longer needed. Provide the sample accessions and state which of them should be removed.

Note that a sample which is left intact can be reused in a new submission, so it is often unnecessary to remove samples when you are only correcting the data attached to them.

Genome Assemblies

An incorrect assembly can be removed so that a corrected version can be submitted. This includes assemblies submitted under the wrong study.

Where the assembly is being corrected rather than withdrawn, consider whether an assembly update is the better route: updates preserve the accession and add a new version. See Update Assembly.

Once the incorrect assembly has been removed, submit the corrected version. If the resubmission produces validation errors, reply on the same helpdesk thread with the error message.

Studies

A study can be removed on its own or together with its associated data. If the study has no linked samples or sequence data, only the study record is cancelled.

State which associated records should be removed and which should be kept. Where sample records are left intact, their accessions remain usable for other submissions.

Can I Request Removal Of Records In Another Webin Account?

That depends on whether you are a named contact on the account that owns them.

• If you are an account contact, you can make the request yourself, in the same way as for your own records.

• If you are not, ENA needs written consent from the account owner before it can act. Alternatively, ask the owner to add you as a contact on the account first, after which you can request the removal directly.

See Collaborating on submissions for how account contacts are managed.

Data brokers cannot suppress data once it has been publicly released, because those data have already been exchanged with the INSDC partners. See Broker Responsibility.

How Do I Release Temporarily Suppressed Data?

If records were temporarily suppressed and are now ready to be public, contact the ENA helpdesk and ask for them to be released.

For run records, provide the corresponding experiment accessions as well as the run accessions, as both are needed to process the request.