Wu, y. | Search bioRxiv

bioRxiv2023

Multi_CycGT: a DL-based multimodal model for membrane permeability prediction of cyclic peptides

As a highly versatile therapeutic modality, cyclic peptides have gained significant attention due to their exceptional binding affinity, minimal toxicity and capacity to target the surface of conventionally "undruggable" proteins. However, the development of cyclic peptides with therapeutic effects by targeting intracellular biological targets has been hindered by the issue of limited membrane permeability. In this paper, we have conducted an extensive benchmarking analysis of a proprietary dataset consisting of 6941 cyclic peptides, employing machine learning and deep learning models. In addition, we propose an innovative multimodal model called Multi_CycGT which combines a Graph Convolutional Network (GCN) and a Transformer to extract 1D and 2D features. These encoded features are then fused for the prediction of cyclic peptide permeability. The cross-validation experiments demonstrate that the proposed Multi_CycGT model achieved the highest level of accuracy on the test set, with an accuracy value of 0.8206 and an AUC value of 0.8650. This paper introduces a pioneering deep learning-based approach that demonstrates enhanced effectiveness in predicting the membrane permeability of cyclic peptides. It also represents the first attempt in this field. We hope that this work will help to accelerate the design of cyclic peptide active drugs in medicinal chemistry and chemical biology applications.

bioinformatics↗

bioRxiv2023

EMBER multi-dimensional spectral microscopy enables quantitative determination of disease- and cell-specific amyloid strains

In neurodegenerative diseases proteins fold into amyloid structures with distinct conformations (strains) that are characteristic of different diseases. However, there is a need to rapidly identify amyloid conformations in situ. Here we use machine learning on the full information available in fluorescent excitation/emission spectra of amyloid binding dyes to identify six distinct different conformational strains in vitro, as well as A{beta} deposits in different transgenic mouse models. Our EMBER (excitation multiplexed bright emission recording) imaging method rapidly identifies conformational differences in A{beta} and tau deposits from Down syndrome, sporadic and familial Alzheimers disease human brain slices. EMBER has in situ identified distinct conformational strains of tau inclusions in astrocytes, oligodendrocytes, and neurons from Picks disease. In future studies, EMBER should enable high-throughput measurements of the fidelity of strain transmission in cellular and animal neurodegenerative diseases models, time course of amyloid strain propagation, and identification of pathogenic versus benign strains. SignificanceIn neurodegenerative diseases proteins fold into amyloid structures with distinct conformations (strains) that are characteristic of different diseases. There is a need to rapidly identify these amyloid conformations in situ. Here we use machine learning on the full information available in fluorescent excitation/emission spectra of amyloid binding dyes to identify six distinct different conformational strains in vitro, as well as A{beta} deposits in different transgenic mouse models. Our imaging method rapidly identifies conformational differences in A{beta} and tau deposits from Down syndrome, sporadic and familial Alzheimers disease human brain slices. We also identified distinct conformational strains of tau inclusions in astrocytes, oligodendrocytes, and neurons from Picks disease. These findings will facilitate the identification of pathogenic protein aggregates to guide research and treatment of protein misfolding diseases.

neuroscience↗