A multi-modal foundation model for state-of-the-art molecular structure prediction of proteins, small molecules, DNA, RNA, and glycosylations.
Chai-1 is a multi-modal foundation model for molecular structure prediction that performs at the state-of-the-art across various benchmarks. It enables unified prediction of proteins, small molecules, DNA, RNA, glycosylations, and other biomolecules, addressing the need for accurate and versatile computational tools in structural biology.
Computational biologists, bioinformaticians, and researchers in drug discovery who require high-accuracy molecular structure prediction for diverse biomolecules.
Developers choose Chai-1 for its state-of-the-art performance, multi-modal capabilities, and support for experimental restraints, offering a unified solution that outperforms specialized models across multiple benchmarks.
Chai-1, SOTA model for biomolecular structure prediction
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Achieves top benchmarks across various molecular types, evidenced by the performance barplot and technical report cited in the README, making it reliable for research.
Unified prediction for proteins, small molecules, DNA, RNA, and glycosylations, eliminating the need for specialized models, as highlighted in the project description.
Allows user-specified inter-chain contacts and covalent bonds to guide folding, a unique feature detailed in the restraints and covalent bond documentation.
Offers CLI, Python API, and a web server for testing, catering to different workflow integration needs, as shown in the installation and running instructions.
Requires specific GPUs like A100 or RTX 4090 with CUDA and bfloat16 support, which can be costly and inaccessible, as noted in the installation section.
Setting up custom MSAs and templates involves understanding file formats like aligned.pqt and m8 files, and managing external servers, adding overhead for users.
MSA generation relies on the ColabFold MMseqs2 server, a shared resource with potential limitations or variability, as admitted in the README details.