AlphaFold Multimer | Folding complexes and multimers using AlphaFold2 | Full Tutorial

In this video we demonstrate how to fold multimers / protein complexes using the AlphaFold2 multimer model on Neurosnap. Easily access AlphaFold2 and other bioinformatic tools online using the free Neurosnap webserver. The Neurosnap platform is designed to make bioinformatics easy for wet lab and non technical biologists.
Access AlphaFold2: neurosnap.ai/s...
Check out Neurosnap: neurosnap.ai/s...
#biology #biochemistry #proteinstructure #molecularbiology #deeplearning #science #biology #bioinformatics #computationalbiology #drugdiscovery #syntheticbiology

[@alexkukreja3509]

What is the best way to delve deeper into scoring the energetics of the PPIs from these predictions? I have run AlphaFold Multimer on a series of homologous proteins and they all produce essentially the same complex (i.e., protein1a binds protein2 using the same interface for binding between protein1b and protein2). Any recommendations for delving deeper into comparing these interactions beyond the general prediction of the most likely interface?

[@NeurosnapInc]

Great question. Follow up steps depend on what exactly you're aiming to do. However, a quick and relatively easy method that we've observed to work is to predict the same complex with other models such as RoseTTAFold2, DiffDock-L on neurosnap (if the ligand is under 50AAs), LightDock, etc... Essentially you want to use a plethora of different tools that perform the same function and then check the similarity between the outputs to see if it's consistent. For instance, if you predict the structure of a binder with its target using five different methods, and all predictions show the same interaction, this strongly suggests the accuracy of the predicted structure. Conversely, if the five predicted structures vary significantly, it would inherently reduce the reliability of the inferred structures.