AlphaFold 601-1200 residues (upload your data)
- Original Image URL: https://github.com/google-deepmind/alphafold
- Expected Completion Time: 240 Minutes
Highlight
Optimizations:
- Maximize CPU and Memory utilization to improve performance
- Use GPU instances only when necessary
- Preposition reference data
Detail
A protein is one or more long chains of amino acids folded into a three-dimensional shape that determines how the protein interacts with other molecules and cell structures. Understanding a protein’s shape is key to understanding a protein’s function and what the consequences are if the protein does not fold correctly. Several diseases, such as cystic fibrosis, Tay-Sachs disease, Marfan syndrome, and some forms of cancer, are linked to misfolded proteins. If the protein’s shape is known, there is the potential to design drugs that can attach to the protein, for example, to block a pathogen such as SARS CoV 2. Synthesizing new proteins for biotech applications, such as making biofuels or degrading waste plastic, depends on knowing the relationship of protein folding to protein function.
Historically, protein-folding studies relied on x-ray crystallography, fluorescence spectroscopy, and other experimental techniques, because model predictions were computationally intractable. This changed in 2018 when Google’s DeepMind group released AlphaFold, its protein-folding model based on artificial intelligence and deep learning of known protein structures.
An implementation of AlphaFold v2 is available under the Apache license on GitHub. MMCloud Air deploys a containerized image of AlphaFold v2.3.2.