Skip to content

Projects

Overview

Collaboration projects are an easy, secure way for data to be shared with specific groups of people. Each project represents a single collaboration and you can have as many as you need. Data stored in a project is automatically only accessible by other project members. You just determine scope of data and the people it needs to be shared with, then we create the project for you with access restricted to specified users.

Please note that only the PI who owns that data may apply for, or request modifications to, a collaboration project.

Why do I need my PI to apply?

We require your PI to apply because they're the data controller on behalf of the University and that responsibility can't be delegated, and also because we need their agreement to our terms and conditions before we can store the data. As much as this may seem annoying, there are some really important things that the PI can do with that data at any time:

  • ask for a new person to be added to the project

  • request ownership of any data be transferred to another project member; especially useful when someone leaves the lab

  • ask for quota to be transferred from one project to another at no cost; this includes creating a new project from unused quota in another, and closing one project to increase the quota of another (assuming the data has been moved)

If you're the PI of a project and would like to do any of these, please contact us via help@imm.ox.ac.uk.

How do collaboration projects work?

Each collaboration project has its own project space in /project. Here's an example of a project with just two members dtooke and emacmahon:

/project/ccbadmin/
                 /dtooke/
                 /emacmahon/
                 /shared/
                 /datashare/

Everything within a project is read-only readable by all other project members, to enable collaboration. Almost all project data access is restricted to project members, ensuring confidentiality. The one exception is the datashare folder, which is intentionally exposed to public Internet. Each project member gets their own directory - only that person can add, remove or modify contents. Each project additionally gets a shared directory, where any project member can write data.

Important safety points

The shared directory by default does not allow unrestricted write access to each others data; only the user that created the data can modify it. This is deliberate. To allow unrestricted write access would be dangerous. Simultaneous file editing runs a risk of unrecoverable data loss, and simultaneous write access to a directory allows anyone to delete all contents by accident. We recommend against trying to "fix" this. If you're absolutely positive that you want unrestricted write access, and you're happy to risk permanent and irreversible data corruption and/or loss, the owner of the data can grant the project group-write permissions with chmod. If you do use chmod, we recommend using syntax chmod [ugo][-+=][rwxXst]+ instead of chmod [0-7]+, e.g. chmod g+w not chmod 660, this avoids accidental unintended permission changes.

How do I join an existing collaboration project?

Please ask the PI in charge of the project to contact us via help@imm.ox.ac.uk and ask for your username to be added.

Successful projects

There's a few things you can do to maximise the likeihood that your projects will be successful. We'd recommend you take the following points into account before you start.

  • First define the set of logically connected data, e.g. a specific research project. Beginning with a list of people then thinking about data they may want to share will risk ambiguous data boundaries that require laborious untangling in future.

  • Create a new collaboration project for each logically separate set of data. Adding unrelated data to an existing project and then adding more people will likely lead to the same problem described above.

  • Please tell us if the data access permissions in your project aren't working for you. Trying to work around an issue on your own is likely to make the problem worse and take longer overall.

  • Please tell us if someone leaves the project and no longer needs access to the data. Leaving accounts with access to data they don't need is a security and privacy risk.

How do I set up a collaboration project?

Setting up a new collaboration project is quick and easy. Please remember that only PIs may request new projects.

  1. Decide the purpose of the project. Each project should be for a single requirement, e.g. data for a lab or research project. If different people need access to different logical data sets, these should be multiple projects. Similarly, if the same people need access to different logical data sets, these should also be multiple projects

  2. Choose a name for the project that isn't already in use in /project/

  3. Decide which user accounts should have access to the data

  4. Decide how much space you think you'll need (in multiples of 1 TB)

  5. Follow these steps.

What about my other directories?

The collaboration projects are an addition to your existing options, not a replacement, meaning that you can have several different locations for your data. Our suggestion for organising your data would be the following:

/home/[a-z]/USERNAME/ - ideal for small private data such as documents or configuration files which are private to you

/project/project1/USERNAME/ - used to store data which is part of project1 but won't typically be used by other people

/project/project1/shared/ - used to store data which is part of project1 and is likely to be used by multiple people