Identification of immune-related tumor antigens and immune subtypes in osteosarcoma.

Purpose: The development of tumor vaccines has become a hot topic in immunotherapy for osteosarcoma (OS); however, more tumor antigens with stronger immunogenicity need to be identified. Methods: We downloaded six sets of gene expression profile data from online databases. The overexpressed genes were analyzed, intersected, and used to calculate the immune infiltration abundance in the TARGET OS dataset based on their expression matrix. Potential tumor antigen genes were identified based on whether they exhibited a high correlation with the antigen-presenting cells (APCs). A total of 1330 immune-related genes (IRGs) from the ImmPort website were retrieved based on their expression, and the Consensus Cluster method was used to obtain immune subtypes of the OS samples. Prognosis, immune microenvironment, and sensitivity to drugs were compared among the immune subtypes. Results: In total, 680 genes were overexpressed in at least two datasets, of which TREM2, TNFRSF12A, and THY1 were positively correlated with different APCs. Based on the expression matrix of 1330 IRGs in TARGET-OS, two immune subtypes, IS1 and IS2, were identified. The prognosis of the IS1 subtype was better than that of IS2, the expression of immune checkpoint (ICP)-related genes was higher in patients with the IS1 subtype, and immune cell infiltration and sensitivity to 16 drugs were generally higher in IS1 subtype patients. Conclusion: We identified three APC-correlated genes that can be considered to code for potential novel tumor antigens for OS vaccines. Two immune subtypes in patients with OS were identified to implement personalized treatments using mRNA vaccines.

Intra-Inter Graph Representation Learning for Protein-Protein Binding Sites Prediction.

Graph neural networks have drawn increasing attention and achieved remarkable progress recently due to their potential applications for a large amount of irregular data. It is a natural way to represent protein as a graph. In this work, we focus on protein-protein binding sites prediction between the ligand and receptor proteins. Previous work just simply adopts graph convolution to learn residue representations of ligand and receptor proteins, then concatenates them and feeds the concatenated representation into a fully connected layer to make predictions, losing much of the information contained in complexes and failing to obtain an optimal prediction. In this paper, we present Intra-Inter Graph Representation Learning for protein-protein binding sites prediction (IIGRL). Specifically, for intra-graph learning, we maximize the mutual information between local node representation and global graph summary to encourage node representation to embody the global information of protein graph. Then we explore fusing two separate ligand and receptor graphs as a whole graph and learning affinities between their residues/nodes to propagate information to each other, which could effectively capture inter-protein information and further enhance the discrimination of residue pairs. Extensive experiments on multiple benchmarks demonstrate that the proposed IIGRL model outperforms state-of-the-art methods.