Christof Seiler
I am a Principal Investigator in the Center of Experimental Rheumatology at University Hospital Zurich and University of Zurich. I taught statistics at Maastricht University from 2018 to 2025. I received my academic training at Stanford University from 2013 to 2018, and at Inria and University of Bern from 2008 to 2012.
Session
11-25
11:25
30min
When hardware decides your biology: What we learned from running PCA on an HPC cluster
Christof Seiler
Principal Component Analysis (PCA) is a standard preprocessing step in genomics data analysis. For large datasets, we run analyses on High-Performance Computing (HPC) clusters and use approximate PCA. In this talk, we will show that different CPU architectures produce different results. We will show code to reproduce the problem and recommend a practical workaround.
Close to the Metal
Room 108