免疫细胞浸润在非小细胞肺癌诊断与预后中的应用

免疫细胞浸润对癌症的诊断与预后有着重要意义。文中收集TCGA数据库已收录的非小细胞肺癌肿瘤与正常组织基因表达数据,利用CIBERSORT工具得到22种免疫细胞占比来评估免疫细胞浸润情况。以22种免疫细胞占比为特征,用机器学习方法构建了非小细胞肺癌肿瘤与正常组织的分类模型,其中随机森林方法构建的模型分类效果AUC=0.987、敏感性0.98及特异性0.84。并且用随机森林方法构建的肺腺癌和肺鳞癌肿瘤组织分类模型效果AUC=0.827、敏感性0.75及特异性0.77。用LASSO回归筛选22种免疫细胞特征,保留8种强相关特征组成的免疫细胞评分结合临床特征构建了非小细胞肺癌预后模型。经评估及验证,预后模型C-index=0.71并且3年和5年的校准曲线拟合良好,可以对预后风险度进行准确预测。本研究基于免疫细胞浸润所构建的分类模型与预后模型,旨在对非小细胞肺癌的诊断与预后研究提供新的策略。     

一种预测个体肿瘤的抗癌药物反应分类计算模型及其应用

目的 不同患者对同一抗癌药物的反应可能不同,了解患者之间对抗癌药物的反应差异对癌症精准医疗具有重大参考价值。方法 高通量测序数据为构建抗癌药物反应分类预测模型提供了强大的数据支撑。针对两大经典数据集癌症细胞百科全书(CCLE)和癌症药物敏感性基因组学数据集(GDSC),本文提出了基于最大相关最小冗余(mRMR)算法和支持向量机(SVM)的计算模型mRMR-SVM。利用基因表达数据,通过方差排序和mRMR算法提取特征基因,借助SVM实现抗癌药物对细胞系的“敏感-抑制”二分类预测。结果 对于CCLE中的22种药物,mRMR-SVM的平均准确率为0.904;对于GDSC中的11种药物,平均准确率为0.851。结论 mRMR-SVM不仅在预测性能方面优于传统的支持向量机、随机森林、深度反应森林、深度神经网络和细胞系-药物复杂网络模型,而且具有良好的泛化能力,对于三类特定组织的抗癌药物反应分类预测也取得了令人满意的结果。此外,mRMR-SVM可以识别与癌症发生发展密切相关的生物标志物。     

浅谈癌症疼痛的临床护理

本文探讨了肿瘤患者疼痛的有效护理措施。对癌症疼痛患者加强非药物止痛、药物止痛护理,心理护理以及控制疼痛的用药指导做了论述。指出加强癌症患者的心理护理及药物指导能有效的控制癌症疼痛,更好的配合治疗。     

肿瘤类器官在药物筛选和个性化用药中的研究进展*

恶性肿瘤是影响人类生命健康的重大疾病之一,药物治疗是常见的治疗手段。近年来,“精准治疗”已经成为肿瘤治疗的趋势。要实现对恶性肿瘤有效、精准的药物治疗,药物筛选模型至关重要。肿瘤类器官是近年来新兴的一种三维细胞模型,具有经长期传代还保留亲本肿瘤的特征和异质性、培养成功率高、周期短和能够高通量筛选药物等优点,已被用于药物筛选、预测患者对治疗的反应以及为个性化用药提供指导等。重点介绍了肿瘤类器官在药物筛选及个性化用药中的研究进展和面临的挑战。     

肿瘤精准医学知识数据库的设计与构建

目的:整合现有前沿的大量而分散的精准医学知识以形成系统完整的知识数据库,为个体组学数据的临床应用提供依据,旨在最终实现基于组学特征的精准用药推荐。方法:采用MySQL数据库管理系统构建数据库,从FDA伴随诊断、NCCN指南、My Cancer Genome、GDSC四大权威医学资源中手动收集精准用药知识,并将原始数据标准化、结构化后以统一的格式存储。结果:成功设计并构建了肿瘤精准医学知识库,目前共收录1 940条精准用药指导,涵盖了基因突变等14种不同类型的组学特征。结论:精准医学知识数据库收录了肿瘤分子组学特征和治疗策略的关联信息,可为临床上个体化治疗方案的制订提供参考依据。数据库的建立为精准医疗临床决策支持系统的开发奠定了基础。     

基于集成模型的小黄鱼越冬群体适宜生境及其环境影响因素

小黄鱼是中韩渔业共同利用鱼种,其跨界洄游习性限制了对越冬场范围的调查和评估,导致对越冬群体适宜栖息地分布缺乏了解。本研究基于越冬期我国自然海域的物种分布点位数据和5个环境数据,运用8个物种分布模型(SDM)分析了小黄鱼越冬场分布范围,采用5折交叉验证,利用受试者工作特征曲线下面积(AUC)评价模型预测性能,并通过加权集成方法构建综合生境模型预测越冬场核心分布位置。结果表明: 出现/未出现数据模型预测准确度普遍高于仅出现模型;在出现/未出现数据模型中,机器学习方法预测准确度高于经典回归模型,支持向量模型(SVM)准确度最高(AUC=0.85),广义线性模型(GLM)准确度最低(AUC=0.73)。集成模型AUC较单一独立模型的准确度有所提升,表明集成模型能有效降低单一独立模型所带来的不确定性,提高模型预测准确度。变量重要性分析结果显示,盐度和温度是决定小黄鱼越冬场地理分布的重要因素,适宜分布区集中在黄海南部外海、东海北部外海和浙江省沿岸海域,而黄海南部沿岸海域和东海中南部外海为不适宜越冬区。研究结果为预测小黄鱼潜在越冬场提供了理论基础,可支撑越冬场渔业资源的空间规划和可持续利用。     

利用组学数据建立针对四种女性癌症的基于机器学习方法的生存预测模型

乳腺癌、宫颈鳞状细胞癌、子宫内膜癌、卵巢癌是女性常见的癌症.由于癌症的恶性发展并缺乏有效的早期诊疗手段,这些癌症已成为当今世界女性患者的头号杀手.为了探索高通量组学数据能否促进癌症患者的预后,本研究利用美国癌症基因组图谱项目中四种女性癌症的1861个样本的临床数据和多维组学数据(包括DNA甲基化、mRNA表达、miRNA表达和基于芯片的蛋白表达组学数据),建立了Cox比例风险模型和随机生存森林模型用来回顾性地预测患者的生存率.本研究发现,在宫颈鳞状细胞癌中,通过整合临床与DNA甲基化及miRNA表达组学数据建立的模型,生存预测能力显著高于仅使用临床数据的模型(一致性指数c-index中位数提高了8.73%～15.03%).本研究虽然验证了有些组学数据对特定癌症生存模型的预测能力有提升作用,但也存在着相对于临床数据,组学数据对模型的预测能力无显著提升的情况.这些结果为系统地开展基于癌症基因组学的生存预测研究及提升临床生存分析的预测准确性提供了有用经验.     

DNA甲基化驱动的转录表达特征作为肝癌预后预测标志物的价值

肝细胞癌(hepatocellular carcinoma,简称肝癌)是最常见的恶性肿瘤之一。DNA甲基化的异常是恶性肿瘤的特征之一,并被发现在肝癌等肿瘤的发生发展中发挥重要作用。为了能为肝癌患者提供新的临床预后预测标志物,本研究首先采用整合组学分析策略在全基因组范围内鉴定与肝癌患者预后相关的DNA甲基化驱动的差异表达基因;然后,采用LASSO (least absolute shrinkage and selection operator)分析建立了10个最优基因组合的预后预测模型。Cox比例风险回归分析显示,在校正临床特征参数后,此预测模型高风险评分与患者不良预后显著相关,表明该模型具有潜在的独立预后价值。受试者工作特征(receiver operating characteristic,ROC)曲线分析显示该风险评分模型在预测患者短期和长期预后方面优于其他已被报道的肝癌预后预测模型。基因集富集分析(gene set enrichment analysis, GSEA)表明,高风险评分与细胞周期和DNA损伤修复通路相关。以上结果表明,本研究构建了一个基于10个DNA甲基化驱动基因的预后风险评分模型,该模型可作为肝癌患者的潜在预后生物标志物,有助于肝癌患者的生存预后评估和治疗策略的指导。     

基于深度学习与多层次信息融合的药物靶标亲和力预测*

药物研发是非常重要但也十分耗费人力物力的过程。利用计算机辅助预测药物与蛋白质亲和力的方法可以极大地加快药物研发过程。药物靶标亲和力预测的关键在于对药物和蛋白质进行准确详细地信息表征。提出一种基于深度学习与多层次信息融合的药物靶标亲和力的预测模型,试图通过综合药物与蛋白质的多层次信息,来获得更好的预测表现。首先将药物表述成分子图和扩展连接指纹两种形式,分别利用图卷积神经网络模块和全连接层进行学习;其次将蛋白质序列和蛋白质K-mer特征分别输入卷积神经网络模块和全连接层来学习蛋白质潜在特征;随后将4个通道学习到的特征进行融合,再利用全连接层进行预测。在两个基准药物靶标亲和力数据集上验证了所提方法的有效性,并与其他已有模型作对比研究。结果说明提出的模型相比基准模型能得到更好的预测性能,表明提出的综合药物与蛋白质多层次信息的药物靶标亲和力预测策略是有效的。     

多因素癌症诊断分类模型的建立和应用

随着基因组学、蛋白质组学技术的运用,对癌症相关分子标志物的筛选,以及基于多个分子标志物的诊断分类模型,成为近年来研究癌症诊断问题的热门途径。本文介绍了如何从基因芯片和质谱等数据库中筛选有诊断作用的分子标志物,以及构建高敏感性、高特异性诊断分类模型的技术路线;并根据在各类肿瘤中的研究实例,分析各分类算法的特点。     

多组学技术及其在生命科学研究中应用概述

随着新一代测序技术、高分辨质谱技术、多组学整合分析方法及数据库的发展,组学技术正从传统的单一组学向多组学技术发展。以多组学驱动的系统生物学研究将带来生命科学研究的新范式。本文简要概述了基因组学、表观基因组学、转录组学,蛋白质组学及代谢组学的进展,重点介绍多组学技术平台的组成和功能,多组学技术的应用现状及在合成生物学及生物医学等领域的应用前景。     

Bayesian Analysis for Uncertainty and Risk in a Groundwater Numerical Model's Predictions

Groundwater modeling typically relies on some hypothesis and approximations of reality, as the real hydrologic systems are far more complex than we can mathematically characterize. This kind of a model's errors cannot be neglected in the uncertainty analysis for a model's predictions in practical issues. As the scale and complexity increase, the associated uncertainties boost dramatically. In this study, a Bayesian uncertainty analysis method for a deterministic model's predictions is presented. The geostatistics of hydrogeologic parameters obtained from site characterization are treated as the prior parameter distribution in the Bayes’ theorem. Then the Markov-Chain Monte Carlo method is used to generate the posterior statistical distribution of the model's predictions, conditional to the observed hydrologic system behaviors. Finally, a series of synthetic examples are given by applying this method to a MODFLOW pumping test model, to test its capability and efficiency in order to assess various sources of the model's prediction uncertainty. The impacts of the model's parameter sensitivity, simplification, and observation errors to predict uncertainty are evaluated, respectively. The results are analyzed statistically to provide deterministic predictions with associated prediction errors. Risk analysis is also derived from the Bayesian results to draw tradeoff curves for decision-making about exploitation of groundwater resources.     

HMMConverter 1.0: a toolbox for hidden Markov models

Hidden Markov models (HMMs) and their variants are widely used in Bioinformatics applications that analyze and compare biological sequences. Designing a novel application requires the insight of a human expert to define the model''s architecture. The implementation of prediction algorithms and algorithms to train the model''s parameters, however, can be a time-consuming and error-prone task. We here present HMMConverter, a software package for setting up probabilistic HMMs, pair-HMMs as well as generalized HMMs and pair-HMMs. The user defines the model itself and the algorithms to be used via an XML file which is then directly translated into efficient C++ code. The software package provides linear-memory prediction algorithms, such as the Hirschberg algorithm, banding and the integration of prior probabilities and is the first to present computationally efficient linear-memory algorithms for automatic parameter training. Users of HMMConverter can thus set up complex applications with a minimum of effort and also perform parameter training and data analyses for large data sets.     

Taming the complexity of large models

There are many complex biological models that fit the data perfectly and yet do not reflect the cellular reality. The process of validating a large model should therefore be viewed as an ongoing mission that refines underlying assumptions by improving low-confidence areas or gaps in the model''s construction.At its most basic, science is about models. Natural phenomena that were perplexing to ancient humans have been systematically illuminated as scientific models have revealed the mathematical order underlying the natural world. But what happens when the models themselves become complex enough that they too must be interpreted to be understood?In 2012, Jonathan Karr, Markus Covert and colleagues at the University of California, San Diego (USA) produced a bold new biological model that attempts to simulate an entire cell: iMg [1]. iMg merges 28 sub-modules of processes within Mycobacterium genitalium, one of the simplest organisms known to man. As a systems biology big-data model, iMg is unique in its scope and is an undeniable paragon of good craft. Because it is probable that this landmark paper will soon be followed by other whole cell models, we feel it is timely to examine this important endeavour, its challenges and potential pitfalls.Building a model requires making many decisions, such as which processes to glaze over and which to reconstruct in detail, how many and what kinds of connections to forge between the model''s constituents, and how to determine values for the model''s parameters. The standard practice has been to tune a model''s parameters and its structure to a best fit with the available data. But this approach breaks down when building a large whole cell model because the number of decisions inflates with the model''s size, and the amount of data required for these decisions to be unequivocal becomes huge. This problem is fundamental, not merely technical, and is rooted in the principle of frugality that underlies all science: Occam''s razor.The problem posed by Occam''s razor is that there are vastly more potential large models that can successfully predict and explain any given body of data than there are small ones. As we can tweak increasingly complex models in an increasing number of ways, we can produce many large models that fit the data perfectly and yet do not reflect the cellular reality. Even if a model fits all the data well, the chance of it happening to be the ‘correct'' model—in other words the one that reflects correctly the underlying cellular architecture and relevant enzymatic parameters—is inversely related to its complexity. A sophisticated large model such as iMg, which has been fitted to many available datasets, will certainly recapture many behaviours of the real system. But it could also recapture many other potentially wrong ones.How do we test a model''s correctness in the sense just mentioned? The intuitive way is to make and test predictions about previously uncharted phenomena. But validating a large biological model is an inherently different challenge than the common practice of “predict, test and validate” customary with smaller ones. Validation using phenotypic ‘emerging'' predictions would require such large amounts of data that it would be highly inefficient and costly at this scale, especially as many of these predictions will turn out to be false leads, with negative results yielding little insight. Rather, the correctness of a whole-cell model is perhaps best validated by using a complementary paradigm: direct testing of the basic decisions that went into the model''s construction. For example, enzymatic rate constants that were fitted in order to make the model behave properly could be experimentally scrutinized for later versions. Performing extensive sensitivity analyses and incorporating known confidence levels of modelling decisions, or harnessing more advanced methods such as ‘active learning'' should all be used in conjunction to determine which parameters to focus on in the future. The process of validating a large model should thus be viewed as an ongoing mission that aims to produce more refined and accurate drafts by improving low-confidence areas or gaps in the model''s construction. Step by step, this paradigm should increase a model''s reliability and ability to make valid new predictions.An open discussion of the potential pitfalls and benefits of building complex biological models could not be timelier, as both the EU and the US have just committed more than a combined 1.4 billion dollars to explicitly model the human brain. Massive data collection and big data analysis are the new norm in most fields, and big models are following closely behind. Their cost, usefulness and application remain open for discussion, but we certainly laud the spirit of the effort. For what is certain is this: only by building these models will we know what usefulness we can attribute to them. Paraphrasing Paul Cezzane, these efforts might be indeed justified and worthy, so long as one is “more or less master of his model”.     

A simple model for the influence of habitat resource availability on lateral clonal spread

Plant clonal spread is ubiquitous and of great interest, owing both to its key role in plant community assembly and its suitability for plant behaviour research. However, mechanisms that govern spreading distance are not well known. Here we link spacer costs and below-ground competition in a simple model of growth in a homogeneous below-ground environment, in which optimal distance between ramets is based on minimizing the sum of these costs. Using this model, we predict a high prevalence of clonal growth that does not employ spacers in resource-poor environments and a nonlinear increase in spreading distance in response to increasing below-ground resource availability. Analysis of database data on clonal growth in relationship to below-ground resource availability revealed that patterns of the spread based on stolons is compatible with the model''s predictions. As expected, model prediction failed for rhizomatous species, where spacer sizes are likely to be selected mainly to play roles other than spread. The model''s simplicity makes it useful as a null model in testing hypotheses about the effects of environmental heterogeneity on clonal spread.     

Modelling seagrass growth and development to evaluate transplanting strategies for restoration

Background and Aims

Seagrasses are important marine plants that are under threat globally. Restoration by transplanting vegetative fragments or seedlings into areas where seagrasses have been lost is possible, but long-term trial data are limited. The goal of this study is to use available short-term data to predict long-term outcomes of transplanting seagrass.

Methods

A functional–structural plant model of seagrass growth that integrates data collected from short-term trials and experiments is presented. The model was parameterized for the species Posidonia australis, a limited validation of the model against independent data and a sensitivity analysis were conducted and the model was used to conduct a preliminary evaluation of different transplanting strategies.

Key Results

The limited validation was successful, and reasonable long-term outcomes could be predicted, based only on short-term data.

Conclusions

This approach for modelling seagrass growth and development enables long-term predictions of the outcomes to be made from different strategies for transplanting seagrass, even when empirical long-term data are difficult or impossible to collect. More validation is required to improve confidence in the model''s predictions, and inclusion of more mechanism will extend the model''s usefulness. Marine restoration represents a novel application of functional–structural plant modelling.     

Regression Analysis for Constraining Free Parameters in Electrophysiological Models of Cardiac Cells

A major challenge in computational biology is constraining free parameters in mathematical models. Adjusting a parameter to make a given model output more realistic sometimes has unexpected and undesirable effects on other model behaviors. Here, we extend a regression-based method for parameter sensitivity analysis and show that a straightforward procedure can uniquely define most ionic conductances in a well-known model of the human ventricular myocyte. The model''s parameter sensitivity was analyzed by randomizing ionic conductances, running repeated simulations to measure physiological outputs, then collecting the randomized parameters and simulation results as “input” and “output” matrices, respectively. Multivariable regression derived a matrix whose elements indicate how changes in conductances influence model outputs. We show here that if the number of linearly-independent outputs equals the number of inputs, the regression matrix can be inverted. This is significant, because it implies that the inverted matrix can specify the ionic conductances that are required to generate a particular combination of model outputs. Applying this idea to the myocyte model tested, we found that most ionic conductances could be specified with precision (R² > 0.77 for 12 out of 16 parameters). We also applied this method to a test case of changes in electrophysiology caused by heart failure and found that changes in most parameters could be well predicted. We complemented our findings using a Bayesian approach to demonstrate that model parameters cannot be specified using limited outputs, but they can be successfully constrained if multiple outputs are considered. Our results place on a solid mathematical footing the intuition-based procedure simultaneously matching a model''s output to several data sets. More generally, this method shows promise as a tool to define model parameters, in electrophysiology and in other biological fields.     

面向药物发现和精准医疗的基因表达谱分析

作为功能基因组学中重要的组成部分,基因表达谱在生物学、医学和药物研发等多个领域发挥着重要作用.特别是随着精准医疗概念的提出,整合多组学数据用于个性化医疗是未来的发展趋势.本文从基因表达谱的基本概念出发,重点介绍面向药物发现的基因表达谱分析方法,即基于关联图谱的方法、基于基因调控网络的方法和基于多组学数据整合的方法.系统整理了各种方法的研究进展,特别是在抗癌药物研发领域的最新进展,为利用基因表达谱数据进行药物研发提供方法借鉴.     

密固达与地舒单抗治疗原发性骨质疏松的成本-效用分析

摘要 目的：评价密固达与地舒单抗治疗原发性骨质疏松的经济性。方法：基于我国卫生体系角度,采用Excel2010软件构建Markov评估模型,利用成本-效用分析的方法评估密固达与地舒单抗治疗原发性骨质疏松的经济性。成本、健康效用值及药物治疗源自已发表的文献。模型循环周期为1年,时效为终生。采用单因素敏感性分析和概率分析评估模型参数变化对结果的影响。结果：地舒单抗用药方案比密固达方案给患者带来0.76质量调整生命年（QALYs）但同时用药成本也高于密固达方案2101.31元,其ICER为2764.88元/QALY。单因素敏感性分析发现药物成本对结果影响较大。概率敏感性分析结果显示,当采用3倍我国2022年人均国内生产总值（GDP）作为意愿支付阈值时,地舒单抗方案更具有经济性。结论：低于3倍我国2022年GDP阈值条件下,地舒单抗治疗原发性骨质疏松更具有经济性。