Reproducible Genome-Wide Analysis With Cloud Computing
summary
Shows that upstream QC choices change GWAS outcomes: one top hit disappears when only the QC pipeline varies. Claims a reproducible pipeline called CLINK built as a versioned virtual appliance around Jupyter and deployed with AWS machine images, containing imaged software and public-data snapshots and Git-controlled workflows. Demonstrates the system on a large case/control study of primary open-angle glaucoma and notes the private neighbor dataset is not bundled.
AI-written summary · about
chapters (9)
- 0:00Start
- 0:30The Reproducibility Crisis
- 0:45Genome-wide Association Studies (GWAS)
- 1:00Quality Control for GWAS is critical
- 1:15Quality Control in the real world
- 1:46Two different QC pipelines: different hits!
- 2:01Reproducibility through DevOps and Cloud
- 4:42Continental ancestry analysis with PCA
- 5:02Conclusions
related talks
23mGenomics and AI on Cloud Foundry
50mCancer Genomics - a biologist and a developer
17mBig data in personalized medicine
42mHow Novel Compute Technology Transforms Life Science Research
32mHow GCP is Powering Next Generation Biomedical Research
21mMachine Learning and Microbiology in the Cloud Through Ray
38mCloud Science
42mBringing Clinical Data to GCP for Automated AI Discoveries