CRII: III: Computational Methods to Explore the Role of Post-transcriptional Regulation in Cancer

该项目将开发计算方法，利用高通量测序数据和分子网络来研究转录后调控在癌症患者中的作用，以提高对癌症的理解，并评估患者的疾病原因。解除基因表达调控是人类肿瘤细胞的一个标志。转录后调控是大多数人类基因关系中的一种普遍机制；它在癌症中的意义才刚刚开始被认识到。该项目的中心假设是，通过考虑转录后调控事件，估计的蛋白质表达可以提供更准确的分子特征来检测复杂的疾病机制，与mRNA表达相比。该项目将研究转录后调控，目的是在不进行大规模蛋白质组学实验的情况下，产生预测蛋白质表达水平变化的计算方法。拟议的研究成果将降低与分析高维基因组图谱互动的障碍，并减少在生物医学研究上花费的时间和成本。新方法将使生物学家和生物医学研究人员能够利用高通量测序数据和生物网络一起进行综合分析，以调查某些癌症类型中转录后调控的影响。这个项目使用多种序列模式分为三个阶段：1)开发先进的机器学习方法，以识别癌症患者和与RNA-Seq数据匹配的正常样本之间mRNA3‘非翻译区(3’UTR)的全基因组选择性多聚腺苷化(APA)事件。2)开发基于生物动机的基于图的学习模型和高效可扩展的算法，利用microRNA-mRNA相互作用网络和3‘UTR-APA事件来估计蛋白质表达水平。3)研究已鉴定的APA和估计的蛋白质表达与典型基因组特征(如mRNA表达、突变和拷贝数变异)相比的预后能力。该项目不仅使用可解释的计算模型来捕捉蛋白质编码基因转录后调控中的因果关系，以更好地理解复杂的人类疾病，而且还为计算机科学研究开发了独特的机器学习方法和优化技术。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This project will develop computational methods to investigate the role of post-transcriptional regulation in cancer patients using both high-throughput sequencing data and molecular networks to improve the understanding of cancer and assess the cause of the disease in a patient. Deregulation of gene expression is a hallmark of the human tumor cells. Post-transcriptional regulation is a pervasive mechanism in the relation of most human genes; its implication in cancer is only beginning to be appreciated. The central hypothesis underlying this project is that by considering post-transcriptional regulation events, estimated protein expressions can provide more accurate molecular signatures to detect complex disease mechanisms, compared to mRNA expressions. This project will study post-transcriptional regulation, with the goal of generating computational methods to predict the changes of protein expression level without doing large-scale proteomics experiments. The outcomes of the proposed research will lower the barriers for interacting with analyzing high-dimensional genomic profiles and cut time and costs spent on biomedical research. The new methods will enable biologists and biomedical researchers to perform comprehensive analysis with high-throughput sequencing data and biological networks together to investigate the impact of post-transcriptional regulation in certain cancer types.This project to use a variety of sequences modalities has three phases: 1) Develop advanced machine learning methods to identify the genome-wide alternative polyadenylation (APA) events in the 3' untranslated region (3'UTR) of the mRNAs between cancer patients and matched normal samples with RNA-Seq data. 2) Develop biologically motivated graph-based learning models and efficient scalable algorithms to estimate the protein expression levels with microRNA-mRNA interaction networks and 3'UTR-APA events. 3) Investigate the prognostic power of the identified APA and estimated protein expressions compared to the canonical genomic features such as mRNA expression, mutation, and copy number variations. The project not only employs interpretable computational models which capture the causal relations in post-transcriptional regulation of protein-coding genes to better understand complex human diseases, but also develops unique machine learning methods and optimization techniques for computer science research.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.