The overview of the software utilized in Simulation is shown in this page. This includes the basic functions of the software, input requirements, output details, and interpretation of the results.
Google Colab provides a dataset that integrates RFdiffusion, Protein MPNN, and AlphaFold2, enabling the design of novel proteins according to workflow:
Protein backbone design with RFdiffusion
Amino acid sequence design with Protein MPNN
Structure prediction with AlphaFold2
These are explained in detail below.
RFdiffusion is an Al model application capable of generating novel protein backbone structures. Particularly, this is handy in term of designing proteins with specific structures or functions, excelling in both 3D structure prediction and design. This model is based on a diffusion model, which incrementally generates and refines protein structures to meet specific objectives. With this approach, the creation of proteins with desired shapes, configurations, amino acid sequences, or physical properties becomes feasible.
Protein MPPN is an AI model to optimize amino acid sequences from a prepared protein structure. Here, this is utilized to design amino acid sequences based on the RFdiffusion-generated protein backbones.
Meanwhile, AlphaFold2 is an AI model to predicts the 3D structure of a protein from amino acid sequences, Thus, 3D structure of the protein which amino acid sequences have been obtained from Protein MPNN can be predicted.
Image on the left shows the example of the use of RFdiffusion.
AlphaFold3 is an advanced AI model, building upon the foundational progress of AlphaFold2, designed for high-precision prediction of biomolecular structures and interactions. This model extends its applicability beyond protein structures to include a broad spectrum of molecular interactions, encompassing DNA, RNA, and small-molecule ligands.
Several data–such as 3D structure of the protein, assesments of structural rigidity and flexibility, and detailed infromation on molecular interactions–is obtainable by inputting an amino acid sequence into AlphaFold3. It is deemed a powerful tool for understanding complex biological system utilizing this expanded functionality.
Inputting an amino acid sequence and pressing "Continue and preview job" will execute the computational process of which the results are shown below.
PyMOL is an open-source molecular visualization tool designed for analyzing the 3D structures of proteins and other biomolecules. It enables the 3D visualization of a wide range of molecules, including proteins, DNA, RNA, and small molecules. Users can highlight specific structural features, such as α-helices and β-sheets, and easily toggle the display of regions of interest, such as binding sites. The software also allows for interactive manipulation of molecular structures, including rotation, zooming, and translation, as well as the measurement of distances and angles between residues. PyMOL is an essential tool for detailed protein visualization and structural analysis.
[1] RFdiffusion v1.1.1 (website)
[2] AlphaFold3 (website)
[3] Joseph L. Watson, David Juergens, Nathaniel R. Bennett, Brian L. Trippe, Jason Yim, Helen E. Eisenach, Woody Ahern, Andrew J. Borst, Robert J. Ragotte, Lukas F. Milles, Basile I. M. Wicky, Nikita Hanikel, Samuel J. Pellock, Alexis Courbet, William Sheffler, Jue Wang, Preetham Venkatesh, Isaac Sappington, Susana Vázquez Torres, Anna Lauko, Valentin De Bortoli, Emile Mathieu, Regina Barzilay, Tommi S. Jaakkola, Frank DiMaio, Minkyung Baek & David Baker (2022) "Broadly applicable and accurate protein design by integrating structure prediction networks and diffusion generative models" bioRxiv