Computational resources for NSF Graduate Research Fellows

The NSF offers computational resources through the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS), including high-performance computing (HPC) and storage. Winners of the National Science Foundation Graduate Research Fellowship (NSF GRF) can request their access by submitting a simple proposal. Credits can then be exchanged for allocations on a variety of partner cluster systems, each of which offers its own specifications and advantages. See mentions of Graduate Students, and NSF Fellows specifically, on the ACCESS policy page. ACCESS has replaced the XSEDE system, though the process for requesting an allocation remains similar; see this post for a simple overview of the application process.

NSF GRF winners are able to serve as a principal investigator on a Discovery Project, meaning you can lead your own project and manage computational resources as a graduate student - which is a great resource system for your project and a training opportunity for you.

Applying for ACCESS

Before submitting your application, you’ll need to define the scope of your research and determine how computational resources will be used. This can include anything from high-throughput data processing to complex simulations. Make sure to clearly state your research objectives and how computational resources are essential to achieving your goals. Emphasizing your role as a PI and how this aligns with your fellowship goals (i.e., a major part of your dissertation) is a great way to strengthen your proposal.

To begin the application process, visit the ACCESS website and create an account. The portal will be your gateway to apply for the resources and manage your allocation.

The application requires:

• A brief public summary of the project
• A one-page main document that includes 1) description of resource needs (e.g., compute resources, parallelization and optimization, storage and tools) and 2) your research objectives
• A “Letter of collaboration from the graduate advisor confirming their awareness of the request and engagement in guiding the computational activity”

Once submitted, your application will be reviewed, and you’ll receive feedback or approval within a few weeks. You can request review by a staff member, which might help you further scope your study (though this is not required).

Using your allocation

Once your application is approved, credits that can be exchanged for the computational resources (on the cluster of your choice) will be added to your account. You’ll receive half your allocated credits at first. To use them, you’ll select the cluster you’d like to use and submit a conversion request. You’ll probably end up primarily using just one cluster, since that’s where your data will be loaded, but you can submit exchange requests across multiple systems.

ACCESS provides access to a variety of computational clusters across the U.S., each optimized for different types of workloads. You’ll need to select the resource that best matches your project’s computational needs. The ACCESS portal provides detailed descriptions of the available resources, including the number of nodes, storage options, and special hardware such as GPUs or specialized memory configurations. You’ll be able to view your project, request additional access, and monitor resource usage through the ACCESS portal. ACCESS also offers technical support, so you can reach out for help with setup or troubleshooting as you scale your project. ACCESS also sends regular emails with training opportunities across a variety of clusters and computational resources across them, which can help you maximize your usage while building your computational skillset across different programs.

After a year, you’ll need to submit a progress report, in which you can request an extension of your allocation as well as the remaining half of your ACCESS credits.

Application example

Here is the text of the public summary I submitted for my project: Our project aims to uncover genetic contributors to irregularities in human reproduction. We will utilize the computational power provided by ACCESS resources to analyze extensive genomic datasets from 20,000 IVF patients, their partners, and 150,000 embryos. This includes investigating the effects of transmission disequilibrium in day-5 embryos, understanding selection processes in early embryonic development, and integrating functional genomics data. To achieve our research goals, we require ACCESS resources for compute power, storage capacity, parallelization, and access to bioinformatics software packages tailored to high-performance computing environments. This project has the potential to advance our understanding of reproductive genetics and improve reproductive health outcomes.

In my one-page document, I gave a two-sentence description of each of the resource needs listed above, with direct connections to my research (for both the specific statisticaly/genetic analyses as well as the necessary software packages). I listed three resarch objectives, each with 3-4 sentences describing the plan and their connection to the overall project.

For the progress report, I detailed the state of the project including the preliminary findings and listed where I’ve shared them thus far (conferences, publications, etc.). I then justfieid the use of the time extension and the remaining credits.