MOTIVATION: MHC Class I protein plays an important role in immunotherapy by presenting immunogenic peptides to anti-tumor immune cells. The repertoires of peptides for various MHC Class I proteins are distinct, which can be reflected by their diverse binding motifs. To characterize binding motifs for MHC Class I proteins, in vitro experiments have been conducted to screen peptides with high binding affinities to hundreds of given MHC Class I proteins. However, considering tens of thousands of known MHC Class I proteins, conducting in vitro experiments for extensive MHC proteins is infeasible, and thus a more efficient and scalable way to characterize binding motifs is needed. RESULTS: We presented a de novo generation framework, coined PepPPO, to characterize binding motif for any given MHC Class I proteins via generating repertoires of peptides presented by them. PepPPO leverages a reinforcement learning agent with a mutation policy to mutate random input peptides into positive presented ones. Using PepPPO, we characterized binding motifs for around 10 000 known human MHC Class I proteins with and without experimental data. These computed motifs demonstrated high similarities with those derived from experimental data. In addition, we found that the motifs could be used for the rapid screening of neoantigens at a much lower time cost than previous deep-learning methods. AVAILABILITY AND IMPLEMENTATION: The software can be found in https://github.com/minrq/pMHC. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

BACKGROUND: It is now clearly evident that cancer outcome and response to therapy is guided by diverse immune-cell activity in tumors. Presently, a key challenge is to comprehensively identify networks of distinct immune-cell signatures present in complex tissue, at higher-resolution and at various stages of differentiation, activation or function. This is particularly so for closely related immune-cells with diminutive, yet critical, differences. RESULTS: To predict networks of infiltrated distinct immune-cell phenotypes at higher resolution, we explored an integrated knowledge-based approach to select immune-cell signature genes integrating not only expression enrichment across immune-cells, but also an automatic capture of relevant immune-cell signature genes from the literature. This knowledge-based approach was integrated with resources of immune-cell specific protein networks, to define signature genes of distinct immune-cell phenotypes. We demonstrate the utility of this approach by profiling signatures of distinct immune-cells, and networks of immune-cells, from metastatic melanoma patients who had undergone chemotherapy. The resultant bioinformatics strategy complements immunohistochemistry from these tumors, and predicts both tumor-killing and immunosuppressive networks of distinct immune-cells in responders and non-responders, respectively. The approach is also shown to capture differences in the immune-cell networks of BRAF versus NRAS mutated metastatic melanomas, and the dynamic changes in resistance to targeted kinase inhibitors in MAPK signalling. CONCLUSIONS: This integrative bioinformatics approach demonstrates that capturing the protein network signatures and ratios of distinct immune-cell in the tumor microenvironment maybe an important factor in predicting response to therapy. This may serve as a computational strategy to define network signatures of distinct immune-cells to guide immuno-pathological discovery.

BACKGROUND: It is of great importance to identify molecular processes and pathways that are involved in disease etiology. Although there has been an extensive use of various high-throughput methods for this task, pathogenic pathways are still not completely understood. Often the set of genes or proteins identified as altered in genome-wide screens show a poor overlap with canonical disease pathways. These findings are difficult to interpret, yet crucial in order to improve the understanding of the molecular processes underlying the disease progression. We present a novel method for identifying groups of connected molecules from a set of differentially expressed genes. These groups represent functional modules sharing common cellular function and involve signaling and regulatory events. Specifically, our method makes use of Bayesian statistics to identify groups of co-regulated genes based on the microarray data, where external information about molecular interactions and connections are used as priors in the group assignments. Markov chain Monte Carlo sampling is used to search for the most reliable grouping. RESULTS: Simulation results showed that the method improved the ability of identifying correct groups compared to traditional clustering, especially for small sample sizes. Applied to a microarray heart failure dataset the method found one large cluster with several genes important for the structure of the extracellular matrix and a smaller group with many genes involved in carbohydrate metabolism. The method was also applied to a microarray dataset on melanoma cancer patients with or without metastasis, where the main cluster was dominated by genes related to keratinocyte differentiation. CONCLUSION: Our method found clusters overlapping with known pathogenic processes, but also pointed to new connections extending beyond the classical pathways.

Two T helper (Th) cell subsets, namely Th1 and Th2 cells, play an important role in inflammatory diseases. The two subsets are thought to counter-regulate each other, and alterations in their balance result in different diseases. This paradigm has been challenged by recent clinical and experimental data. Because of the large number of genes involved in regulating Th1 and Th2 cells, assessment of this paradigm by modeling or experiments is difficult. Novel algorithms based on formal methods now permit the analysis of large gene regulatory networks. By combining these algorithms with in silico knockouts and gene expression microarray data from human T cells, we examined if the results were compatible with a counter-regulatory role of Th1 and Th2 cells. We constructed a directed network model of genes regulating Th1 and Th2 cells through text mining and manual curation. We identified four attractors in the network, three of which included genes that corresponded to Th0, Th1 and Th2 cells. The fourth attractor contained a mixture of Th1 and Th2 genes. We found that neither in silico knockouts of the Th1 and Th2 attractor genes nor gene expression microarray data from patients with immunological disorders and healthy subjects supported a counter-regulatory role of Th1 and Th2 cells. By combining network modeling with transcriptomic data analysis and in silico knockouts, we have devised a practical way to help unravel complex regulatory network topology and to increase our understanding of how network actions may differ in health and disease.