Interpretable deep learning models for translational medicine

用于转化医学的可解释深度学习模型

基本信息

批准号：
10579895
负责人：
XINGHUA LU
金额：
$ 31.37万
依托单位：
UNIVERSITY OF PITTSBURGH AT PITTSBURGH
依托单位国家：
美国
项目类别：
财政年份：
2015
资助国家：
美国
起止时间：
2015-04-01 至 2025-03-31
项目状态：
未结题

来源：
https://reporter.nih.gov/project-details/10579895
关键词：
Affect Algorithms Antineoplastic Agents Big Data Biological Cancer Patient Cancer cell line Cell model Cell physiology Cells Chemicals Data Disease Event Gene Expression Genetic Genetic Transcription Grain Human Immune Evasion Immunotherapy Individual Information Theory Intervention Knowledge Learning Libraries Malignant Neoplasms Maps Messenger RNA Methodology MicroRNAs Mining Modeling Molecular Monitor Nature Network-based Organoids Outcome Paper Pathologic Pathway interactions Patients Phenotype Physiological Publishing Research Research Personnel Side Signal Pathway Signal Transduction Signaling Molecule Structure System Systems Biology Techniques Technology The Cancer Genome Atlas Training Translational Research United States National Institutes of Health Yeasts biological systems cancer cell cancer therapy cell behavior data modeling deep field survey deep learning deep learning algorithm deep learning model design drug sensitivity experience genome-wide innovation inquiry-based learning insight learning algorithm learning strategy machine learning algorithm machine learning method novel pharmacologic pre-clinical precision medicine precision oncology predicting response prevent response single-cell RNA sequencing success theories tool transcription factor transcriptome transcriptomics translational applications translational impact translational medicine transmission process treatment response tumor tumor microenvironment

项目摘要

Understanding the state of cellular signaling systems provides insights to how cells behave under physiological and pathological conditions. Cellular signaling systems are organized as hierarchy (cascade) and signals of a molecular is often compositionally encoded to control cellular processes, such as gene expression. This project aims to develop advanced deep learning models (DLMs) to simulate cellular signaling systems based on gene expression data. In last 3 years, the project has made significant progresses, but the challenges remain. Importantly, contemporary DLMs behave as “black boxes”, in that it is difficult to interpret how signals are encoded and how to interpret which signal a hidden node represent in a DLM. This black-box nature prevents researchers from gaining biological insights using DLMs, even though these models can be much superior in modeling data than other types of models in many tasks, e.g., predicting drug sensitivity of cancer cells. In this competitive renewal, we propose to develop novel DLMs and innovative inference algorithms to train “interpretable” DLMs and apply them in translational research. The proposed research is innovative and of high significance in several perspectives: 1) Our novel DLMs and algorithms take advantage of big data resulting from systematic chemical/genetic perturbations of cellular signaling machinery, so that we can use the perturbation condition as side information to reveal how signals are encoded in a DLM. 2) We integrate principles of causal inference and information theory with deep learning method to make DLMs interpretable. As results, that researchers can gain mechanistic insights from such models. 3) Innovative application of interpretable DLMs will advance translational research. For example, we will train interpretable DLMs to model cellular signaling at the level of single cells and use this information investigate inter-cellular interactions among cells in tumor microenvironment to shed light on immune evasion mechanisms of cancers. We will also use information derived from interpretable DLMs to predict cancer cell drug sensitivity. We anticipate that our study will bring forth significant advances not only in deep learning methodology but also in precision medicine.

了解细胞信号系统的状态，为细胞在生理下的表现提供了见解和病理状况。细胞信号系统被组织为层次结构（级联）和信号分子通常被合成编码，以控制细胞过程，例如基因表达。这项目旨在开发先进的深度学习模型（DLM）来模拟基于蜂窝信号系统关于基因表达数据。在过去的三年中，该项目取得了重大进展，但是挑战保持。重要的是，当代DLM的行为为“黑匣子”，因为很难解释信号编码以及如何解释隐藏节点在DLM中表示的信号。这种黑色盒子的性质防止研究人员使用DLM获得生物学见解，即使这些模型可能很重要在许多任务中，比其他类型的模型都优越，例如预测癌症的药物敏感性细胞。在这种竞争性更新中，我们建议开发新颖的DLM和创新的推论算法训练“可解释”的DLM，并将其应用于翻译研究。拟议的研究是创新的，从几个角度来看：1）我们的新型DLM和算法利用了大数据由细胞信号机械的系统化学/遗传扰动产生，以便我们可以使用扰动条件作为侧面信息，以揭示信号如何在DLM中编码。 2）我们整合因果推理和信息理论的原则，采用深度学习方法使DLM可解释。结果，研究人员可以从此类模型中获得机理见解。 3）创新应用可解释的DLM将推进转化研究。例如，我们将训练可解释的DLM进行建模单个细胞水平的细胞信号传导并使用此信息研究细胞间相互作用在肿瘤微环境中的细胞中，可以阐明癌症的免疫进化机制。我们也会使用可解释DLM的信息来预测癌细胞药物敏感性。我们预计我们的研究将不仅在深度学习方法论，而且还将在精确医学上带来重大进步。

项目成果

期刊论文数量（17）

专著数量（0）

科研奖励数量（0）

会议论文数量（0）

专利数量（0）

Toward multimodal signal detection of adverse drug reactions.

DOI：
10.1016/j.jbi.2017.10.013
发表时间：
2017-12
期刊：
Journal of biomedical informatics
影响因子：
4.5
作者：
Harpaz R;DuMouchel W;Schuemie M;Bodenreider O;Friedman C;Horvitz E;Ripple A;Sorbello A;White RW;Winnenburg R;Shah NH
通讯作者：
Shah NH

Inferring causal molecular networks: empirical assessment through a community-based effort.

DOI：
10.1038/nmeth.3773
发表时间：
2016-04
期刊：
Nature methods
影响因子：
48
作者：
Hill SM;Heiser LM;Cokelaer T;Unger M;Nesser NK;Carlin DE;Zhang Y;Sokolov A;Paull EO;Wong CK;Graim K;Bivol A;Wang H;Zhu F;Afsari B;Danilova LV;Favorov AV;Lee WS;Taylor D;Hu CW;Long BL;Noren DP;Bisberg AJ;HPN-DREAM Consortium;Mills GB;Gray JW;Kellen M;Norman T;Friend S;Qutub AA;Fertig EJ;Guan Y;Song M;Stuart JM;Spellman PT;Koeppl H;Stolovitzky G;Saez-Rodriguez J;Mukherjee S
通讯作者：
Mukherjee S

An exact algorithm for finding cancer driver somatic genome alterations: the weighted mutually exclusive maximum set cover problem.

DOI：
10.1186/s13015-016-0073-9
发表时间：
2016
期刊：
Algorithms for molecular biology : AMB
影响因子：
0
作者：
Lu S;Mandava G;Yan G;Lu X
通讯作者：
Lu X

A Novel Bayesian Framework Infers Driver Activation States and Reveals Pathway-Oriented Molecular Subtypes in Head and Neck Cancer.

DOI：
10.3390/cancers14194825
发表时间：
2022-10-03
期刊：
CANCERS
影响因子：
5.2
作者：
Liu, Zhengping;Cai, Chunhui;Ma, Xiaojun;Liu, Jinling;Chen, Lujia;Lui, Vivian Wai Yan;Cooper, Gregory F.;Lu, Xinghua
通讯作者：
Lu, Xinghua

A signal-based method for finding driver modules of breast cancer metastasis to the lung.