Yucheng Guo

Abstract Chronic inflammation is a serious risk factor for cancer; however, the routes from inflammation to cancer are poorly understood. On the basis of the processes implicated by frequently mutated genes associated with inflammation and cancer in three organs (stomach, colon, and liver) extracted from the Gene Expression Omnibus, The Cancer Genome Atlas, and Gene Ontology databases, we present a multiscale model of the long-term evolutionary dynamics leading from inflammation to tumorigenesis. The model incorporates cross-talk among interactions on several scales, including responses to DNA damage, gene mutation, cell-cycle behavior, population dynamics, inflammation, and metabolism-immune balance. Model simulations revealed two stages of inflammation-induced tumorigenesis: a precancerous state and tumorigenesis. The precancerous state was mainly caused by mutations in the cell proliferation pathway; the transition from the precancerous to tumorigenic states was induced by mutations in pathways associated with apoptosis, differentiation, and metabolism-immune balance. We identified opposing effects of inflammation on tumorigenesis. Mild inflammation removed cells with DNA damage through DNA damage-induced cell death, whereas severe inflammation accelerated accumulation of mutations and hence promoted tumorigenesis. These results provide insight into the evolutionary dynamics of inflammation-induced tumorigenesis and highlight the combinatorial effects of inflammation and metabolism-immune balance. This approach establishes methods for quantifying cancer risk, for the discovery of driver pathways in inflammation-induced tumorigenesis, and has direct relevance for early detection and prevention and development of new treatment regimes. Cancer Res; 77(22); 6429–41. ©2017 AACR.

Abstract Large-scale pretrained models have become foundation models leading to breakthroughs in natural language processing and related fields. Developing foundation models in life science for deciphering the “languages” of cells and facilitating biomedical research is promising yet challenging. We developed a large-scale pretrained model scFoundation with 100M parameters for this purpose. scFoundation was trained on over 50 million human single-cell transcriptomics data, which contain high-throughput observations on the complex molecular features in all known types of cells. scFoundation is currently the largest model in terms of the size of trainable parameters, dimensionality of genes and the number of cells used in the pre-training. Experiments showed that scFoundation can serve as a foundation model for single-cell transcriptomics and achieve state-of-the-art performances in a diverse array of downstream tasks, such as gene expression enhancement, tissue drug response prediction, single-cell drug response classification, and single-cell perturbation prediction.

Tumorigenesis is a complex process that is driven by a combination of networks of genes and environmental factors; however, efficient approaches to identifying functional networks that are perturbed by the process of tumorigenesis are lacking. In this study, we provide a comprehensive network-based strategy for the systematic discovery of functional synergistic modules that are causal determinants of inflammation-induced tumorigenesis. Our approach prioritizes candidate genes selected by integrating clinical-based and network-based genome-wide gene prediction methods and identifies functional synergistic modules based on combinatorial CRISPR-Cas9 screening. On the basis of candidate genes inferred de novo from experimental and computational methods to be involved in inflammation and cancer, we used an existing TGFβ1-induced cellular transformation model in colonic epithelial cells and a new combinatorial CRISPR-Cas9 screening strategy to construct an inflammation-induced differential genetic interaction network. The inflammation-induced differential genetic interaction network that we generated yielded functional insights into the genes and functional module combinations, and showed varied responses to the inflammation agents as well as active traditional Chinese medicine compounds. We identified opposing differential genetic interactions of inflammation-induced tumorigenesis: synergistic promotion and suppression. The synergistic promotion state was primarily caused by deletions in the immune and metabolism modules; the synergistic suppression state was primarily induced by deletions in the proliferation and immune modules or in the proliferation and metabolism modules. These results provide insight into possible early combinational targets and biomarkers for inflammation-induced tumorigenesis and highlight the synergistic effects that occur among immune, proliferation, and metabolism modules. In conclusion, this approach deepens the understanding of the underlying mechanisms that cause inflammation to potentially increase the cancer risk of colonic epithelial cells and accelerate the translation into novel functional modules or synergistic module combinations that modulate complex disease phenotypes.

INTRODUCTION: Breast cancer, one of the most common health threats to females worldwide, has always been a crucial topic in the medical field. With the rapid development of digital pathology, many scholars have used AI-based systems to classify breast cancer pathological images. However, most existing studies only stayed on the binary classification of breast lesions (normal vs tumor or benign vs malignant), far from meeting the clinical demand. Therefore, we established a multi-class classification system of breast digital pathology images based on AI, which is more clinically practical than the binary classification system. METHODS: In this paper, we adopted a two-stage architecture based on deep learning method and machine learning method for the multi-class classification (normal tissue, benign lesion, ductal carcinoma in situ, and invasive carcinoma) of breast digital pathological images. RESULTS: The proposed approach achieved an overall accuracy of 86.67% at patch-level. At WSI-level, the overall accuracies of our classification system were 88.16% on validation data and 90.43% on test data. Additionally, we used two public datasets, the BreakHis and BACH, for independent verification. The accuracies our model obtained on these two datasets were comparable to related publications. Furthermore, our model could achieve accuracies of 85.19% on multi-classification and 96.30% on binary classification (non-malignant vs malignant) using pathology images of frozen sections, which was proven to have good generalizability. Then, we used t-SNE for visualization of patch classification efficiency. Finally, we analyzed morphological characteristics of patches learned by the model. CONCLUSION: The proposed two-stage model could be effectively applied to the multi-class classification task of breast pathology images and could be a very useful tool for assisting pathologists in diagnosing breast cancer.