Originally published at: Accelerating Single Cell Genomic Analysis using RAPIDS | NVIDIA Technical Blog
The human body is made up of nearly 40 trillion cells, of many different types. Recent advances in experimental biology have made it possible to explore the genetic material of single cells. With the birth of this new field of single-cell genomics, scientists can now probe the DNA and RNA of individual cells in the…
1 Like
Very cool application!
Our lab has extensively refactored our code to incorporate RAPIDS libraries to accelerate our single-cell analyses.
I recently re-implemented the widely-used single-cell phenotyping algorithm PhenoGraph (GitHub - jacoblevine/PhenoGraph: Subpopulation detection in high-dimensional single-cell data) using a combination of cuML, cuGraph, cupy, and cupyx.sparse. The GPU-based implementation (Erik Burlingame / grapheno · GitLab) yields an orders-of-magnitude boost in speed over the CPU-based implementation, without sacrificing cell cluster quality.
Keep up the great work, RAPIDS team!
We’re excited to accelerate single-cell research with RAPIDS and enable better, faster science. If you have any questions or comments, or are interested in applying RAPIDS to your single-cell research, please let us know.
1 Like
Great to see that you’re seeing speedups using RAPIDS. I run the RAPIDS blog. Would you be interested in contributing a guest blog on your use case? I’d be happy to collaborate.
1 Like
Hi, @mbeaumont. Absolutely! I believe I have a Jupyter notebook that might serve as a good starting point for a blog post.
Can you reach out to me over email? mbeaumont@nvidia.com
Hello @jwitsoe
When loading data to the GPU for your testing was that done with only one core of the CPU? During my walking though your guide (great guide by the way), I noticed that when I ran htop. I was thinking the performance would be better if more cores loaded the GPU with data. Is this being done by one core the expected behavior?