The NucleicNet is hosted on our webserver (http://www.cbrc.kaust.edu.sa/NucleicNet/)
Protein-RNA interaction is ubiquitous in cells and serves as the main mechanism for post-transcriptional regulation. Base-dominant interaction and backbone-dominant interaction categorize the two main modes of the way RNA interacting with proteins. Despite the advances in experimental technologies and computational methods to capture protein-RNA interactions, estimating binding preference of RNA backbone constituents and different bases on any location of a given protein surface is beyond the capacity of existing techniques. Here we show that these attributes can be predicted from the local physicochemical characteristics of the protein structure surface, by leveraging the power of deep learning.
The NucleicNet is hosted on our webserver (http://www.cbrc.kaust.edu.sa/NucleicNet/). Here, we distribute a version that operates on Linux machines (Centos 7 /Ubuntu >= 16). For instructions on installation and usage, please refer to our structured Wiki page (https://github.com/NucleicNet/NucleicNet/wiki). Examples of PDB format files are stored in the "GridData" folder of the current directory. Expected outputs (except files exceeding 100 MB in size) are stored at the "Example" folder (https://github.com/NucleicNet/NucleicNet/tree/master/Example).
Lam, J.H., Li, Y., Zhu, L. et al. A deep learning framework to predict binding preference of RNA constituents on protein surface. Nat Commun 10, 4941 (2019) doi:10.1038/s41467-019-12920-0