- Speaker： Dr. Yang Ni, the Department of Statistics, Texas A&M University.
- Title： MAIB-Talk-012: Manifold Learning and Artificial Intelligence -Causal Graphical Models for Discovering Gene Regulations
- Date： Saturday, 10:00 pm US East time, 04/08/2023
- Date： Sunday, 10:00 am Beijing time, 04/09/2023
- Zoom ID： 933 1613 9423
- Zoom PWD： 416262 -Zoom: https://uwmadison.zoom.us/meeting/register/tJcudu-prTIuGNda1MsF8PKyRQlnGn06TP2E
Title: Causal Graphical Models for Discovering Gene Regulations
Abstract: Dr. Yang Ni will present several causal graphical models for discovering gene regulations from observational genomic data in an exploratory fashion. Our methods are specifically tailored to common features of genomic data including high level of noise, high skewness, zero-inflation, sample heterogeneity, feedback loops, and presence of unmeasured confounders. Our theories show that causal structure is identifiable under all the presented causal graphical models with purely observational data. I will provide intuition as to why causality is identifiable under different scenarios and demonstrate the practical utility using multiple real datasets with known causal structure.
摘要：我将介绍几种因果图模型，用于探索性地从观测到的基因组数据中发现基因调控。我们的方法针对基因组数据的常见特征进行了特别定制，包括高噪声水平、高偏斜度、零膨胀、样本异质性、反馈环路和未测到的混淆因素存在等。我们的理论表明，在所有呈现的因果图模型中，都可以通过纯观测数据确定因果结构。我将提供因果性在不同情境下可识别的直觉，并使用多个已知因果结构的真实数据集展示实际效用。
Bio: Dr. Yang Ni is an Assistant Professor in the Department of Statistics, Texas A&M University. I am the Co-Director of the Single Cell Data Science Core, a Research Affiliate at the Texas A&M Institute of Data Science (TAMIDS), and the Co-Director of the Center for Statistical Bioinformatics. I enjoy working on problems that are at the intersection of statistics, artificial intelligence, philosophy, biology, and health. Recently, I became obsessed with endurance sports (swim, bike, and run). https://callmespring.github.io/
简介：Dr. Yang Ni是德克萨斯A&M大学统计学系的助理教授。我是单细胞数据科学核心的联合主任，德克萨斯A&M数据科学研究所（TAMIDS）的研究联合会员，以及统计生物信息学中心的联合主任。我喜欢处理统计学、人工智能、哲学、生物学和健康等交叉学科的问题。最近，我对耐力运动（游泳、自行车和跑步）产生了极大的兴趣。
Background
Causal Graphical Models (CGMs) are a powerful tool for discovering gene regulations from observational genomic data. CGMs are graphical representations of causal relationships between variables, and they can be used to model the complex interactions between genes and other biological factors. By analyzing these models, researchers can identify the underlying mechanisms that control gene expression, which can help them develop new therapies and treatments for a range of diseases.
CGMs are especially useful for analyzing genomic data because they can handle the high levels of noise, skewness, and zero-inflation that are commonly found in this type of data. They can also account for sample heterogeneity, feedback loops, and the presence of unmeasured confounders. In addition, CGMs are able to identify causal structure from purely observational data, which means that researchers can use them to make predictions and test hypotheses without conducting expensive and time-consuming experiments.
By using CGMs to analyze genomic data, researchers can gain new insights into the complex relationships between genes and other biological factors. This can help them identify new targets for drug development and develop more effective treatments for diseases. In addition, CGMs can be used to identify biomarkers that can be used to predict disease progression and response to treatment, which can help improve patient outcomes and reduce healthcare costs.
Overall, CGMs are a valuable tool for discovering gene regulations from observational genomic data, and they have the potential to revolutionize the field of genomics and lead to new breakthroughs in the diagnosis and treatment of disease.
该领域目前正在快速发展，并涉及多个学科，包括统计学、计算机科学、生物学和医学等。近年来，随着大规模基因组数据的不断涌现和技术的不断进步，因果推断和因果图模型在基因调控和疾病机制研究中得到越来越广泛的应用。
一些最新的研究成果表明，因果图模型能够成功地发现基因调控的因果结构，并在疾病机制研究中取得了重要的进展。此外，研究人员还在不断改进因果图模型的性能，以更好地处理基因组数据的特征，例如高噪声水平、高偏斜度、零膨胀、样本异质性、反馈环路和未测到的混淆因素存在等。
总的来说，该领域的发展前景广阔，未来还将有更多的工作集中在基因调控的因果推断和疾病机制研究上，并在更多的领域中得到广泛应用。
虽然因果图模型在基因调控和疾病机制研究中的应用取得了一些进展，但是该领域还存在许多问题需要解决。以下是一些急需解决的问题：
如何在处理高维基因组数据时提高因果推断的效率和准确性？
如何有效地处理基因组数据的非线性关系和高度复杂的交互作用？
如何应对样本异质性和未测到的混淆因素的影响？
如何设计更加准确和可靠的实验来验证因果推断的结果？
如何利用因果图模型来更好地理解基因调控和疾病机制的复杂性？
如何利用因果图模型来指导药物研发和治疗策略的制定？
解决这些问题需要不断地改进因果图模型和开发新的算法和工具，同时也需要更加深入地了解基因调控和疾病机制的本质。
如何确定基因调控的因果结构？是否存在其他方法来验证所提出的因果结构？
基于这些因果图模型，我们如何解释基因调控的机制？是否有助于理解疾病的发生和治疗？
在处理高噪声水平、高偏斜度、零膨胀、样本异质性、反馈环路和未测到的混淆因素存在等基因组数据特征时，这些因果图模型的表现如何？是否存在更好的方法来处理这些数据特征？
除了基因调控之外，这些因果图模型是否适用于其他领域，例如社会科学或医学研究？
采用纯观测数据确定因果结构的方法是否具有普适性？在实际应用中是否需要考虑其他因素或使用其他方法进行验证？
因果图模型可以帮助识别药物作用机制和影响药物疗效的生物学因素。首先，可以使用因果图模型分析药物对基因调控的影响，帮助确定药物的作用靶点和影响的信号通路。其次，因果图模型可以分析药物对生物体整体状态的影响，帮助预测药物对疾病的治疗效果和不良反应。最后，因果图模型可以帮助设计合理的临床试验，确定患者的纳入标准和治疗方案，从而提高治疗效果和减少不必要的风险和成本。
因此，利用因果图模型来指导药物研发和治疗策略的制定具有广泛的应用前景和重要意义。
