Cyclic peptides represent a highly promising class of biopharmaceutical scaffolds. The screening of cyclic peptides against protein targets can be greatly facilitated using computational approaches, especially molecular docking. However, it remains a crucial challenge to accurately predict protein-cyclic peptide (P-cp) interactions employing scoring functions of molecular docking. End-point approaches, such as molecular mechanics generalized Born surface area (MM/GBSA) and molecular mechanics Poisson-Boltzmann surface area (MM/PBSA), provide theoretically more robust frameworks than conventional scoring functions, but their reliability in predicting binding affinities and discriminating native-like binding poses for P-cp complexes remains poorly quantified. Herein, we comprehensively assessed the predictive abilities of MM/PBSA(GBSA) in scoring binding affinities of P-cp complexes and re-ranking their binding poses. The binding affinity scoring ability of MM/PBSA(GBSA) was assessed on a carefully curated dataset consisting of 50 complexes involving P-cp binding affinities, and their re-ranking capability was evaluated on another dataset consisting of the decoys of 81 P-cp complexes. Based on these assessments, we proposed a two-step workflow for predicting P-cp binding affinities. First, we employed the assessed optimal re-ranking method to select the top-1 binding pose; second, we estimated the binding affinity based on the selected top-1 pose using the assessed optimal scoring method. Our proposed workflow, which requires only 3 s for each prediction, achieves binding affinity predictions with a Rp of -0.732 when compared to experimental values, which is twice as high as that of AutoDock CrankPep (Rp = -0.316). This study emphasizes the necessity of using fine-tuned MM/PBSA(GBSA) methods for predicting P-cp interactions.

Therapeutic peptides have proven to have great pharmaceutical value and potential in recent decades. However, methods of AI-assisted peptide drug discovery are not fully explored. To fill the gap, we propose a target-aware peptide design method called PPFlow, based on conditional flow matching on torus manifolds, to model the internal geometries of torsion angles for the peptide structure design. Besides, we establish a protein-peptide binding dataset named PPBench2024 to fill the void of massive data for the task of structure-based peptide drug design and to allow the training of deep learning methods. Extensive experiments show that PPFlow reaches state-of-the-art performance in tasks of peptide drug generation and optimization in comparison with baseline models, and can be generalized to other tasks including docking and side-chain packing.

Myopia has become a major public health problem affecting the eye health of our citizens, especially teenagers. Fundus images can be obtained non-invasively and can be used to monitor and follow up on the progress in high myopia. However, with the development of artificial intelligence, it is still difficult to establish a computer-aided diagnosis model for high myopia with young children as research objects, mainly because 1) it is very difficult to collect labeled fundus images, and there is no large amount of such data that can be freely accessed; 2) hard samples which have clinical significance for population screening and diagnosis are rare and indistinguishable. To solve these problems, we propose a high myopia detection model on fundus images based on curriculum learning. We design a dual-curriculum generation module, which aims to use the expert model to endow the curriculum with new training indicators so that the student model can gradually and robustly identify hard samples. Compared with the baseline, our framework significantly improves the convergence speed of the training process and achieves the best performance during testing. Experimental results on a high myopia fundus images dataset show that our framework provides efficient and accurate detection and outperforms other methods.

Accurate cooperative localization (CL) is a crucial requirement for multi-AUVs operation. However, the Velocity Uncertainty (VU) measured by DVL (Doppler Velocity Logger) in certain environments makes it challenging for master-slaver CL. A single beacon-aided CL positioning method based on Maximum Correntropy Criterion (MCC) is proposed in this paper. Firstly, a velocity correction method is presented to deal with the problem of VU. Then a circular intersection solving method relying on time transmission delay is adopted to achieve slaver AUV's location. In which a judging rule is designed to distinguish mirror solutions in circular intersection equations. Simulation results verify that the proposed algorithm can achieve a high localization accuracy with velocity uncertainty.