Best Practices:

• Look to campus high-performance computing resources as a first start, if available.
• If not, look to the Computer Science department for cloud-based computing courses.
• Understand how others in the same field of research use cloud-based computing, and leverage their experience whenever possible.
• Participate in training courses for neuroscience data in the cCoud.
• Work with the Cloud vendor to understand options, costs, and programs to support your work.
   

Things to Avoid:

• Do not assume you know what resources are available without checking. You need to do due diligence to check on resources.
• Do not be limited by the computing power of your laptop. There are many resources that will get you computing for free or a low cost.
• Do not invest in solutions that do not have a clear track record and evidence of some longevity.

Value Set Definitions: 

• Low: Researcher has access to few institutional resources. Researchers who would label themselves as low on this dimension should consider carefully whether they have the time and resources needed to develop the necessary expertise in order to use cloud-based resources for their projects.
• Mixed: Good neuroimaging expertise, but little institutional computer or data science support; or good computational expertise and resources but little neuroimaging expertise.
• High: Good neuroimaging expertise and strong institutional computer and data science support using Cloud Computing.
   

Value of Use Case Example:

Mixed - Jordan’s institute has a high-performance computing center for use by investigators with expertise in using containerized (i.e., pre-packaged) computing tools, and a data science center with “cloud office hours”, i.e., a dedicated help resource that can answer her questions about using the Cloud. Further, there is at least one investigator in another department that regularly uses cloud-based computing resources. However, no one in Jordan’s department does so. Jordan’s institution has been involved in a number of large-scale clinical studies that involved data sharing of clinical data, but has not done so with neuroimaging data in previous studies.

Discussion of Use Case:

Based on the resources available to Jordan, she should reach out to colleagues with specific expertise in cloud-based neuroimaging or consider taking a training course that provides training in cloud-based use of neuroimaging tools.