Background/Objectives: Accurately predicting protein–ligand binding affinity is essential in drug discovery for identifying effective compounds. While existing sequence-based machine learning models for binding affinity prediction have shown potential, they lack accuracy and robustness in pattern recognition, which limits their generalizability across diverse and novel binding complexes. To overcome these limitations, we developed GNNSeq, a novel hybrid machine learning model that integrates a Graph Neural Network (GNN) with Random Forest (RF) and XGBoost. Methods: GNNSeq predicts ligand binding affinity by extracting molecular characteristics and sequence patterns from protein and ligand sequences. The fully optimized GNNSeq model was trained and tested on subsets of the PDBbind dataset. The novelty of GNNSeq lies in its exclusive reliance on sequence features, a hybrid GNN framework, and an optimized kernel-based context-switching design. By relying exclusively on sequence features, GNNSeq eliminates the need for pre-docked complexes or high-quality structural data, allowing for accurate binding affinity predictions even when interaction-based or structural information is unavailable. The integration of GNN, XGBoost, and RF improves GNNSeq performance by hierarchical sequence learning, handling complex feature interactions, reducing variance, and forming a robust ensemble that improves predictions and mitigates overfitting. The GNNSeq unique kernel-based context switching scheme optimizes model efficiency and runtime, dynamically adjusts feature weighting between sequence and basic structural information, and improves predictive accuracy and model generalization. Results: In benchmarking, GNNSeq performed comparably to several existing sequence-based models and achieved a Pearson correlation coefficient (PCC) of 0.784 on the PDBbind v.2020 refined set and 0.84 on the PDBbind v.2016 core set. During external validation with the DUDE-Z v.2023.06.20 dataset, GNNSeq attained an average area under the curve (AUC) of 0.74, demonstrating its ability to distinguish active ligands from decoys across diverse ligand–receptor pairs. To further evaluate its performance, we combined GNNSeq with two additional specialized models that integrate structural and protein–ligand interaction features. When tested on a curated set of well-characterized drug–target complexes, the hybrid models achieved an average PCC of 0.89, with the top-performing model reaching a PCC of 0.97. GNNSeq was designed with a strong emphasis on computational efficiency, training on 5000+ complexes in 1 h and 32 min, with real-time affinity predictions for test complexes. Conclusions: GNNSeq provides an efficient and scalable approach for binding affinity prediction, offering improved accuracy and generalizability while enabling large-scale virtual screening and cost-effective hit identification. GNNSeq is publicly available in a server-based graphical user interface (GUI) format.

Previous epitope-based cancer vaccines have focused on analyzing a limited number of mutated epitopes and clinical variables preliminarily to experimental trials. As a result, relatively few positive clinical outcomes have been observed in epitope-based cancer vaccines. Further efforts are required to diversify the selection of mutated epitopes tailored to cancers with different genetic signatures. To address this, we developed the first version of AutoEpiCollect, a user-friendly GUI software, capable of generating safe and immunogenic epitopes from missense mutations in any oncogene of interest. This software incorporates a novel, machine learning-driven epitope ranking method, leveraging a probabilistic logistic regression model that is trained on experimental T-cell assay data. Users can freely download AutoEpiCollectGUI with its user guide for installing and running the software on GitHub. We used AutoEpiCollect to design a pan-cancer vaccine targeting missense mutations found in the proto-oncogene PIK3CA, which encodes the p110ɑ catalytic subunit of the PI3K kinase protein. We selected PIK3CA as our gene target due to its widespread prevalence as an oncokinase across various cancer types and its lack of presence as a gene target in clinical trials. After entering 49 distinct point mutations into AutoEpiCollect, we acquired 361 MHC Class I epitope/HLA pairs and 219 MHC Class II epitope/HLA pairs. From the 49 input point mutations, we identified MHC Class I epitopes targeting 34 of these mutations and MHC Class II epitopes targeting 11 mutations. Furthermore, to assess the potential impact of our pan-cancer vaccine, we employed PCOptim and PCOptim-CD to streamline our epitope list and attain optimized vaccine population coverage. We achieved a world population coverage of 98.09% for MHC Class I data and 81.81% for MHC Class II data. We used three of our predicted immunogenic epitopes to further construct 3D models of peptide-HLA and peptide-HLA-TCR complexes to analyze the epitope binding potential and TCR interactions. Future studies could aim to validate AutoEpiCollect’s vaccine design in murine models affected by PIK3CA-mutated or other mutated tumor cells located in various tissue types. AutoEpiCollect streamlines the preclinical vaccine development process, saving time for thorough testing of vaccinations in experimental trials.

Accurately predicting protein–ligand binding affinity is an important step in the drug discovery process. Deep learning (DL) methods have improved binding affinity prediction by using diverse categories of molecular data. However, many models rely heavily on interaction and sequence data, which impedes proper learning and limits performance in de novo applications. To address these limitations, we developed a novel graph neural network model, called StructureNet (structure-based graph neural network), to predict protein–ligand binding affinity. StructureNet improves existing DL methods by focusing entirely on structural descriptors to mitigate data memorization issues introduced by sequence and interaction data. StructureNet represents the protein and ligand structures as graphs, which are processed using a GNN-based ensemble deep learning model. StructureNet achieved a PCC of 0.68 and an AUC of 0.75 on the PDBBind v.2020 Refined Set, outperforming similar structure-based models. External validation on the DUDE-Z dataset showed that StructureNet can effectively distinguish between active and decoy ligands. Further testing on a small subset of well-known drugs indicates that StructureNet has high potential for rapid virtual screening applications. We also hybridized StructureNet with interaction- and sequence-based models to investigate their impact on testing accuracy and found minimal difference (0.01 PCC) between merged models and StructureNet as a standalone model. An ablation study found that geometric descriptors were the key drivers of model performance, with their removal leading to a PCC decrease of over 15.7%. Lastly, we tested StructureNet on ensembles of binding complex conformers generated using molecular dynamics (MD) simulations and found that incorporating multiple conformations of the same complex often improves model accuracy by capturing binding site flexibility. Overall, the results show that structural data alone are sufficient for binding affinity predictions and can address pattern recognition challenges introduced by sequence and interaction features. Additionally, structural representations of protein–ligand complexes can be considerably improved using geometric and topological descriptors. We made StructureNet GUI interface freely available online.

Personalized cancer vaccines are a key strategy for training the immune system to recognize and respond to tumor-specific antigens. Our earlier software release, AutoEpiCollect 1.0, was designed to accelerate the vaccine design process, but the identification of tumor-specific genetic variants remains a manual process and is highly burdensome. In this study, we introduce AutoEpiCollect 2.0, an improved version with integrated genetic analysis capabilities that automate the identification and prioritization of tumorigenic variants from individual tumor samples. AutoEpiCollect 2.0 connects with RNA sequencing and cross-references the resulting RNAseq data for efficient determination of cancer-specific and prognostic gene variants. Using AutoEpiCollect 2.0, we conducted two case studies to design personalized peptide vaccines for two distinct cancer types: cervical squamous cell carcinoma and breast carcinoma. Case 1 analyzed five cervical tumor samples from different stages, ranging from CIN1 to cervical cancer stage IIB. CIN3 was selected for detailed analysis due to its pre-invasive status and clinical relevance, as it is the earliest stage where patients typically present symptoms. Case 2 examined five breast tumor samples, including HER2-negative, ER-positive, PR-positive, and triple-negative subtypes. In three of these breast samples, the same epitope was identified and was synthesized by identical gene variants. This finding suggests the presence of shared antigenic targets across subtypes. We identified the top MHC class I and class II epitopes for both cancer types. In cervical carcinoma, the most immunogenic epitopes were found in proteins expressed by HSPG2 and MUC5AC. In breast carcinoma, epitopes with the highest potential were derived from proteins expressed by BRCA2 and AHNAK2. These epitopes were further validated through pMHC-TCR modeling analysis. Despite differences in cancer type and tumor subtype, both case studies successfully identified high-potential epitopes suitable for personalized vaccine design. The integration of AutoEpiCollect 2.0 streamlined the variant analysis workflow and reduced the time required to identify key tumor antigens. This study demonstrates the value of automated data integration in genomic analysis for cancer vaccine development. Furthermore, by applying RNAseq in a standardized workflow, the approach enables both patient-specific and population-level vaccine design, based on statistically frequent gene variants observed across tumor datasets. AutoEpiCollect 2.0 is freely available as a website based tool for user to design vaccine.