Biology subjects

Rademaker, D. T.

Publications and source records attributed to Rademaker, D. T..

6 recordsLinked to original sources

bioRxiv2026

Expanding the options for therapeutic exon skipping as a future treatment for USH2A-associated disease by 3D structural modeling of newly formed hybrid domains

Usher syndrome, the leading cause of hereditary deaf-blindness affecting approximately 1 in 15,000 individuals worldwide, is currently still untreatable. Antisense oligonucleotide-based exon skipping has shown significant therapeutic promise for USH2A-associated retinal dysfunction. Selection of (combinations of) exons suitable for therapeutic exon skipping within the fibronectin type 3 (FN3) domain-encoding region of USH2A currently requires that skipped exons exactly align with complete protein domains. However, only few exon combinations meet this criterion, which significantly restricts the therapeutic potential of this strategy. Our study addresses this limitation by incorporating AlphaFold2 structural modelling into the exon skipping target selection pipeline. Following this adjusted framework, we can predict exon skipping combinations that allow remaining domain fragments to form structurally viable hybrid domains. As a proof-of-concept, we examined and confirmed the functionality of usherin{Delta}exon54-58 that contains a hybrid FN3 domain, using zebrafish as a model. This highligts the potential of the newly developed paradigm for identifying exon skipping targets with potential therapeutic relevance. Our results emphasize the value of structural modeling in identifying new therapeutic exon skipping targets, aiming to improve precision, efficiency, applicability, and cost-effectiveness in the development of genetic therapies for hereditary diseases such as Usher syndrome.

bioinformatics↗

bioRxiv2025

SwiftMHC: A High-Speed Attention Network for MHC-Bound Peptide Identification and 3D Modeling

Identifying tumor peptides that bind patient MHC proteins and elicit immune responses is central to immunotherapy, yet progress remains limited to a handful of well-studied alleles. Structure-based methods generalize better than sequence-only models but are constrained by the high computational cost of 3D modeling. We present SwiftMHC, the fastest structure-based framework for peptide-MHC (pMHC) modeling and binding affinity prediction. SwiftMHC predicts peptide-MHC binding in 0.009 sec/case in batch mode on a single A100 GPU--nearly an order of magnitude faster than leading sequence-based methods such as NetMHCpan 4.1 (0.081 sec/case)--while performing competitively in predictive accuracy. In addition to affinity estimation, SwiftMHC generates all-atom 3D pMHC structures and achieves a median C-RMSD of 1.32 [A] against X-ray benchmarks, matching or better than the accuracy of state-of-the-art approaches (e.g., AlphaFold with fine-tuning) but running thousands of times faster (excluding disk-writing time). These results demonstrate the power of task-specific AI trained on physics-derived synthetic data to overcome the scarcity of experimental structures. Optimized for HLA-A*02:01 9-mers but readily extensible to other alleles, SwiftMHC enables rapid and accurate identification of peptides distinct from self at T-cell-exposed surfaces. This capability may expand immunotherapy targets, improve the safety of TCR-based therapies, and accelerate the development of next-generation cancer immunotherapies.

immunology↗

bioRxiv2024

Bidirectional Peptide Conformation Prediction for MHC Class II using PANDORA

Recent discoveries have transformed our understanding of peptide binding in Major Histocompatibility Complex (MHC) molecules, showing that peptides, for some MHC class II alleles, can bind in a reverse orientation (C-terminus to N-terminus) and can still effectively activate CD4+ T cells. These finding challenges established concepts of immune recognition and suggests new pathways for therapeutic intervention, such as vaccine design. We present an updated version of PANDORA, which, to the best of our knowledge, is the first tool capable of modeling reversed-bound peptides. Modeling these peptides presents a unique challenge due to the limited structural data available for these orientations in existing databases. PANDORA has overcome this challenge through integrative modeling using algorithmically reversed peptides as templates. We have validated the new PANDORA feature through a series of experiments, achieving an average backbone binding-core L-RMSD value of 0.63 [A]. Notably, it maintained low RMSD values even when using templates from different alleles and peptide sequences. Our results suggest that PANDORA will be an invaluable resource for the immunology community, aiding in the development of targeted immunotherapies and vaccine design. AvailabilitySource code and data is freely available at https://github.com/X-lab-3D/PANDORA; Contact: Li.Xue@radboudumc.nl

immunology↗

bioRxiv2023

Improving generalizability for MHC-binding peptide predictions through structure-based geometric deep learning

The interaction between peptides and major histocompatibility complex (MHC) molecules is pivotal in autoimmunity, pathogen recognition and tumor immunity. Recent advances in cancer immunotherapies demand for more accurate computational prediction of MHC-bound peptides. We address the generalizability challenge of MHC-bound peptide predictions, revealing limitations in current sequence-based approaches. Our structure-based methods leveraging geometric deep learning (GDL) demonstrated promising improvement in generalizability across unseen MHC alleles. Further, we tackle data efficiency by introducing a self-supervised learning approach on structures (3D-SSL). Without being exposed to any binding affinity data, our 3D-SSL outperforms sequence-based methods trained on [~]90 times more datapoints. Finally, we demonstrate the resilience of structure-based GDL methods to biases in binding data on an Hepatitis B virus vaccine immunopeptidomics case study. This proof-of-concept study highlights structure-based methods potential to enhance generalizability and data efficiency, with important implications for data-intensive fields like T-cell receptor specificity predictions, paving the way for enhanced comprehension and manipulation of immune responses.

bioinformatics↗

bioRxiv2023

Synergizing Geometric Deep Learning and Data-Centric Methods for Improved Protein Structural Alignment

MotivationStructures are replacing the role of sequences. Traditional bioinformatics research focused on sequences because they were easily obtained. Advances in techniques like cryo-electron microscopy, molecular modeling, docking algorithms, and structure prediction software have shifted the focus to structures. Given the importance of deep learning in many of these breakthroughs, it makes sense to also explore how it can modernize classic bioinformatics tools. However, empirical findings have shown that machine learning based methods have many pitfalls resulting in overoptimistic conclusions, including data leakage between test and training data. Thus, there is a need for new innovations to make neural networks more intelligible. ResultsWe have developed vanGOGH, a geometric deep learning-based structural alignment approach that performs on par with the state-of-the-art without ever having been trained on a pair of naturally found homologs. We adopted a data-centric approach to address deep learning and data limitations by augmenting protein templates into synthetic homologs for training. Our method allowed us to supplement homolog data by knowledge-driven augmentation, self-learning of relevant structural features by supervised examples and protein alignment that is competitive with state-of-the art methods. AvailabilityGNN framework: https://github.com/DeepRank/deeprank-core/tree/main/deeprankcore ContactLi.Xue@radboudumc.nl

bioinformatics↗

bioRxiv2022

Quantifying the deformability of malaria-infected red blood cells using deep learning trained on synthetic cells

Several haematologic diseases, including malaria, diabetes, and sickle cell anaemia, result in a reduced red blood cell deformability. This deformability can be measured using a microfluidic device with channels of varying width. Nevertheless, it is challenging to algorithmically recognise large numbers of red blood cells and quantify their deformability from image data. Deep learning has become the method of choice to handle noisy and complex image data. However, it requires a significant amount of labelled data to train the neural networks. By creating images of cells and mimicking noise and plasticity in those images, we generate synthetic data to train a network to detect and segment red blood cells from video-recordings, without the need for manually annotated labels. Using this new method, we uncover significant differences between the deformability of RBCs infected with different strains of Plasmodium falciparum, providing clues to the variation in virulence of these strains.

microbiology↗