XCloud-MoDern: An Artificial Intelligence Cloud for Accelerated NMR Spectroscopy

Zi Wang,Di Guo,Zhangren Tu,Yihui Huang,Yirong Zhou,Jian Wang,Liubin Feng,Donghai Lin,Yongfu You,Tatiana Agback,Vladislav Yu. Orekhov,Xiaobo Qu
2020-01-01
Abstract:For accelerated multi-dimensional NMR spectroscopy, non-uniform sampling is a powerful approach but requires sophisticated algorithms to reconstruct undersampled data. Here, we first devise a high-performance deep learning framework (MoDern), which shows astonishing performance in robust and high-quality reconstruction of challenging multi-dimensional protein NMR spectra and reliable quantitative measure of the metabolite mixture. Remarkably, the few trainable parameters of MoDern allowed the neural network to be trained on solely synthetic data while generalizing well to experimental undersampled data in various scenarios. Then, we develop a novel artificial intelligence cloud computing platform (XCloud-MoDern), as a reliable, widely-available, ultra-fast, and easy-to-use technique for highly accelerated NMR. All results demonstrate that XCloud-MoDern contributes a promising platform for further development of spectra analysis.
What problem does this paper attempt to address?