FAQ

Q: Can I use a sample in multiple projects?

A: An ICA sample contains multiple files related to one or more bio samples (eg, multiple lanes). You can link samples to multiple projects; however, a file can only be added to one sample.

Q: What does it mean for files and samples to be linked to a project?

A: Files and samples linked to a project allow the data to be used when launching pipelines within that project.

Q: I have a private docker image that I want to upload to a tool and only work on in my project. Does anyone in workgroup have access to docker images?

A: Yes, docker images are available to the entire domain when added to the Docker Repository in ICA.

Q: When creating a tool, if I have my own CWL doc, can I just import it instead of building the tool using the UI?

A: Yes, you can take your CWL docs and paste it in instead of configuring a tool via the UI. However, you might need to change the namespace. Best practice is to try it first locally using cwltool.

Q: Can I add more than one domain admin?

A: Yes, see the BaseSpace Sequence Hub documentation for information on adding domain admins.

Q: What are the file size limits when uploading/downloading data through the user interface compared to the CLI?

A: File uploads via the UI are limited to 5GB. File uploads via the CLI and Connector have no size limit (uploads >5GB use multipart upload).

Q: Since the inputs/outputs have already been configured in tools, why do I need to reconfigure in the pipeline?

A: Tools define the inputs and outputs available for that specific Tool, and pipelines connect the inputs and outputs to other steps/Tools or the initial inputs/final output files. Pipelines can be shared across projects and tools, so you might want the behavior to differ.

Q: Why can I specify a compute type for each tool when building a pipeline and again when running the pipeline?

A: When building the pipeline, you can specify the default compute type for each tool. When running the pipeline, you can override the default to be used for that specific analysis.

Q: If I upload a docker, which workgroups within the domain have access?

A: All workgroups have access. Docker images in ICA are available at the domain level.

Q: How come there are run time differences between my ICA analysis compared to local high performance clusters?

A: There can be several reasons why ICA run time differences will differ with a local cluster:

  1. Network Latency: ICA run time may differ compared to a local cluster due to data transfer between the client and server, which can vary in time due to the network being shared and data traveling across servers.

  2. Virtualization Overhead: ICA uses virtualization, which can result in a performance differences compared to a local cluster, where the hardware is dedicated to the compute task.

  3. Shared Resources: ICA is a shared resource platform, which means that the resources available to a user are not dedicated.

    1. Instance type: The type of instance being spun up can affect how long it takes to start up. Some instance types have more resources and take longer to initialize, while others are optimized for quick startup times.

    2. Availability zones: If the desired availability zone is already at capacity or has limited resources available, it may take longer for Kubernetes to spin up instances.

Last updated