ProtTrans and multi-window scanning convolutional neural networks for the prediction of protein-peptide interaction sites.

This study delves into the prediction of protein-peptide interactions using advanced machine learning techniques, comparing models such as sequence-based, standard CNNs, and traditional classifiers. Leveraging pre-trained language models and multi-view window scanning CNNs, our approach yields significant improvements, with ProtTrans standing out based on 2.1 billion protein sequences and 393 billion amino acids.

Generation of a plasmonic radially polarized vector beam with linearly polarized illumination.

Polarization state of a wave field can be manipulated through the plasmonic metasurface consisting of orthogonal nanoslit pairs; the output polarization angle is independent of the incident linearly polarized light and is highly dependent on the orientations of nanoslit pairs. We combine the Archimedes spiral with the nanoslit pairs to compensate for the Pancharatnam-Berry (PB) phase induced by the orientation of nanoslits, as well as achieve the radially polarized vector beam (RPVB) under the illuminations of different linearly polarized lights. Experiments are performed to successfully realize the RPVB, and the results are in excellent agreement with the numerical simulations.