Open arschat opened 3 months ago
Data Processing Pipeline Support
from diagram & textConsent
Criteria, current text:
The current HCA data policy requires fully open access datasets. Any managed access data cannot be accepted at this time.
There is no mention of living donors here, therefore, no rephrasing to showcase responsibility of data owner.
Refactor Data Coordination Platform
to Data Platform
Update Open and Controlled access field - add bold/ remove strikedout
Open and Controlled access
Depending on the data access documents supporting the study and the respective data, datasets eligible for the Human Cell Atlas would either be
Open access
- where the raw data is available for open access by the scientific community and can be made publicly, e.g. in a public archive
- where the data owner can take written responsibility to share and is aware that the data will be openly shared through the HCA Data Portal without access controls in place to monitor who has accessed the data and for which research purpose
Controlled access
- where data with personally identifiable information without consent to be shared publicly. E.g raw sequences, is under controlled access. This type of access usually has to be granted by a nominated DAC (Data Access Committee)
The Human Cell Atlas Data Coordination Platform accepts datasets featuring both data access types, with an important note on the data and metadata files required for each type:Open access datasets - accepted data and metadata:
- Raw sequencing files (fastq, bam)- Donor and sample metadata- Analysis files (raw/normalised matrices; cell type annotations; spatial embeddings)Controlled access datasets - accepted data and metadata:
- Sample metadata (no donor-identifiable metadata)- Analysis files (raw/normalised matrices; cell type annotations; spatial embeddings)
Update Contribution checklist - add bold/ remove strikedout
Contribution checklist
To contribute data to the Human Cell Atlas Data Coordination Platform, follow this checklist:
- Make sure your dataset conforms to the suitability criteria above.
- Get informed about the data access type for your dataset (Open access, Managed
Controlledaccess, or a mix of both).E.g. Data from living donors is always Controlled access.- Sign the DCA agreement if required
- Prepare the data files and metadata for your dataset, based on the requirements for your data access type.
- Fill in the HCA dataset questionnaire to inform us of dataset details.
- Contact HCA Data Coordination Platform data wrangling team to get you started.
Assess data availability
Check the type of consent:
- Open: suitable.
Managed/Controlled access: Needs to at least have publicly accessible count matrices. If there is no shareable data the dataset should be down prioritised or possibly considered ineligible.- Managed/ Controlled access: if data are not public, we could send email to suggest the managed access route (link to managed access SOPs). If raw data is not available for open or managed access publication in Data Portal, we could publish with public available contributor generated matrices.
Additionally, GDPR is in consideration; if the dataset contains samples from living donors,
no externally obtained data or metadata can be added. We consider externally obtained data any type of information that is not obtained from public sources (e.g. the manuscript or archives such as GEO).we should ask the data owner to take (written) responsibility to share and inform them that the data will be openly shared through the HCA Data Portal without access controls in place to monitor who has accessed the data and for which research purposeCheck that the sequence data is available (similar to the above). If the sequence data is fully available in fastq format, this is suitable. If the sequence data cannot be found or is not in fastq format, then the dataset can still be included if there are publicly sharable contributor generated matrices. If there is no shareable data in the form of fastq files or a gene expression matrix, the dataset should be down prioritised or possibly considered ineligible.
The clause of publishing with only analysis files, might need to be changed after discussions with HCA execs, since CxG are the place for the count matrices of the source datasets.
@idazucchi check our gdpr doc and our communication doc -
We need to update the diagram :
Wranglers are only allowed to take the information that the authors have made already of public domain. if the authors provide more metadata that what's already publicly available and also provide written consent for its public release then the additional metadata can be wrangled as open access
no edit needed for now but we will need to update it after the managed access pilot is done
PR for Contributor Guide ebi-ait/ingest-ui#196 PR for Data suitability & GDPR guidelines in Wrangler's SOP #1273 For Data portal update, we need to investigate the process for updates in the DataBiosphere/data-portal repo.
In order to do a PR in the DataBiosphere/data-portal, we would need to create ticket that describes changes needed, make PRs and ask Dave's review. The changes proposed are: PR 1: Removing Pipeline support from docs
Data Processing Pipeline Support
from diagram & text Data Processing and Results
page
PR 2: Add processed files optionProcessed data files
after Unprocessed data files
option
PR 3: Add Managed Access optionCriteria - Consent
field in diagram & text HCA exec office will review data.humancellatlas.org (HCA-OPS point -> needs a thorough review)
Other PRs need review, and asked Gabs & Ida.
Asking execs if there is a published DCA pdf that will be updated, so we could link to that in Contributor Guide
Until DCA is finalised, there won't be a linkable published DCA pdf. We can share wrangler's email if people want to ask about DCA.
Description of the task:
After legal discussions about MA, we discovered that if we get written consent from contributor (email) that they want (have consent) to publish potentially person-identifiable metadata, we can proceed with publishing. This is a little bit different from what we thought previously (that we are not allowed to publish any data from living donors if it was not published before), and it is not reflected in the docs that we hold.
We have identified places where such changes might be good: 1 - https://data.humancellatlas.org/contribute/contributing-data-suitability 2 - https://contribute.data.humancellatlas.org/guide 3 - https://ebi-ait.github.io/hca-ebi-wrangler-central/SOPs/Triage/dataset_suitability_SOP.html
We discussed that docs 1,2 might need to be identical, and doc 3 is lower priority.
Acceptance criteria for the task: