Project title
Computational Structural Biology Node
Collaborators and funding
Contact(s)
- Keiran Rowell - Scientific Officer - UNSW, Structural Biology Facility - k.rowell@unsw.edu.au
- Kate Michie - Chief Scientist - UNSW, Structural Biology Facility - k.michie@unsw.edu.au
Project description and aims
The Structural Biology Node will explore software and hardware efficiencies in the current deep learning revolution in computational structural biology. Pawsey’s architecture provides a unique testing ground for structural biology software on high performance computing (HPC). These findings will be shared with local and national HPC facilities, the steering committee, and scientific advisory board in order to formulate best-practice in this new style of compute for biomolecular structures, and drive widespread adoption by biochemical/medical researchers in Australia.
Aims:
- Benchmarking AlphaFold variants (e.g.
OpenFold,FastFold) and optimising their use on HPC facilities - Apply protein generative AI (e.g.
RFDiffusion,EvoDiff,ProteinMPNN) for generation of novel protein designs that can be used in fundamental biology and therapeutic research - Validating compilation of CUDA code of the above software onto AMD GPUs.
How is ABLeS supporting this work?
This work is supported through the production bioinformatics scheme provided by ABLeS. The supports includes unlimited temporary storage on scratch, 5 TB permanent storage and 100 KSUs GPU allocation per quarter.
Expected outputs enabled by participation in ABLeS
This project enables the Structural Biology Node to be responsive to the deep learning developments in computational structural biology. The repurposing and application of AI methods to biological problems is producing pre-print code at a rapid rate, many of these methods find eventual publication in top-rank journals. This is the beginning of software explosion in this ecosystem, and so we will be continually validating new code on the leading-edge for wide use by the biological research community. Findings will be shared to the wider community and may ouput publications in technical proceedings.
These details have been provided by project members at project initiation. For more information on the project, please consult the contact(s) or project links above.