Mission: While not synonymous with Precision Medicine, Precision Genomics plays a crucial role within its broader scope alongside fields like Precision Psychiatry and Precision Imaging. As a research lab, the Turner Lab is dedicated to advancing Precision Genomics by tackling five key challenges(in the above image) that currently limit its application for everyone.
Resulted Success: In the Turner Lab, success is measured by published papers, the types of grants received, and whether trainees are learning from this research and becoming the next generation to impact the field in new ways.
Expert Insights: “We’ve done lots of AlphaFold calculations on the RIS. They would have cost millions of dollars if we had run them in Google Cloud. Running them on the RIS made an impossible problem possible because our lab does not currently have millions of dollars to run these computations in the cloud. We do, however, have a nice RIS setup to get the work done.” – Tychele Turner, Assistant Professor of Genetics. – Dr. Tychele Turner, Assistant Professor of Genetics.
“We created a new tool, 3D-Clump, to look at clustering of variants on protein structures, and we were aided by being able to make AlphaFold predictions on the RIS.” – Dr. Tychele Turner, Assistant Professor of Genetics.
Developed Collaboration: Hare And Tortoise, HAT, are two de novo variant callers developed by the Tychele N. Turner, Ph.D. Lab with the help of Research Infrastructure Services at WashU. Hare, as seen in Ng et al. 2022, uses the software Parabricks, v4.0.0-1, by NVIDIA, that leverages GPUs to accelerate variant calling, specifically for Haplotyecaller GATK 4.2.0 and DeepVariant v1.4.0. Tortoise uses freely available, open-source versions of these variant callers. The lab uses GLnexus to form family-level joint-genotyped files that can be run through their custom de novo variant filter workflow.
| By the Number: RIS Resources at Work |
|---|
| 2.5 PB Active storage |
| 1 PB of Archive storage |
| 272 CPUs, 4 GPUs, and 1 Condo server |
| 4,395,828 jobs consumed 13,391,704 CPU core hours and 216,467 GPU hours |