基于图自编码器和协同训练预测miRNA[]与疾病的关联

近年来,越来越多的生物学实验研究表明,microRNA (miRNA)在人类复杂疾病的发展中发挥着重要作用。因此,预测miRNA与疾病之间的关联有助于疾病的准确诊断和有效治疗。由于传统的生物学实验是一种昂贵且耗时的方式,于是许多基于生物学数据的计算模型被提出来预测miRNA与疾病的关联。本研究提出了一种端到端的深度学习模型来预测miRNA-疾病关联关系,称为MDAGAC。首先,通过整合疾病语义相似性,miRNA功能相似性和高斯相互作用谱核相似性,构建miRNA和疾病的相似性图。然后,通过图自编码器和协同训练来改善标签传播的效果。该模型分别在miRNA图和疾病图上建立了两个图自编码器,并对这两个图自编码器进行了协同训练。miRNA图和疾病图上的图自编码器能够通过初始关联矩阵重构得分矩阵,这相当于在图上传播标签。miRNA-疾病关联的预测概率可以从得分矩阵得到。基于五折交叉验证的实验结果表明,MDAGAC方法可靠有效,优于现有的几种预测miRNA-疾病关联的方法。     

GCAEMDA: Predicting miRNA-disease associations via graph convolutional autoencoder

microRNAs (miRNAs) are small non-coding RNAs related to a number of complicated biological processes. A growing body of studies have suggested that miRNAs are closely associated with many human diseases. It is meaningful to consider disease-related miRNAs as potential biomarkers, which could greatly contribute to understanding the mechanisms of complex diseases and benefit the prevention, detection, diagnosis and treatment of extraordinary diseases. In this study, we presented a novel model named Graph Convolutional Autoencoder for miRNA-Disease Association Prediction (GCAEMDA). In the proposed model, we utilized miRNA-miRNA similarities, disease-disease similarities and verified miRNA-disease associations to construct a heterogeneous network, which is applied to learn the embeddings of miRNAs and diseases. In addition, we separately constructed miRNA-based and disease-based sub-networks. Combining the embeddings of miRNAs and diseases, graph convolutional autoencoder (GCAE) was utilized to calculate association scores of miRNA-disease on two sub-networks, respectively. Furthermore, we obtained final prediction scores between miRNAs and diseases by adopting an average ensemble way to integrate the prediction scores from two types of subnetworks. To indicate the accuracy of GCAEMDA, we applied different cross validation methods to evaluate our model whose performances were better than the state-of-the-art models. Case studies on a common human diseases were also implemented to prove the effectiveness of GCAEMDA. The results demonstrated that GCAEMDA was beneficial to infer potential associations of miRNA-disease.     

Laplacian normalization and bi-random walks on heterogeneous networks for predicting lncRNA-disease associations

Background

Evidences have increasingly indicated that lncRNAs (long non-coding RNAs) are deeply involved in important biological regulation processes leading to various human complex diseases. Experimental investigations of these disease associated lncRNAs are slow with high costs. Computational methods to infer potential associations between lncRNAs and diseases have become an effective prior-pinpointing approach to the experimental verification.

Results

In this study, we develop a novel method for the prediction of lncRNA-disease associations using bi-random walks on a network merging the similarities of lncRNAs and diseases. Particularly, this method applies a Laplacian technique to normalize the lncRNA similarity matrix and the disease similarity matrix before the construction of the lncRNA similarity network and disease similarity network. The two networks are then connected via existing lncRNA-disease associations. After that, bi-random walks are applied on the heterogeneous network to predict the potential associations between the lncRNAs and the diseases. Experimental results demonstrate that the performance of our method is highly comparable to or better than the state-of-the-art methods for predicting lncRNA-disease associations. Our analyses on three cancer data sets (breast cancer, lung cancer, and liver cancer) also indicate the usefulness of our method in practical applications.

Conclusions

Our proposed method, including the construction of the lncRNA similarity network and disease similarity network and the bi-random walks algorithm on the heterogeneous network, could be used for prediction of potential associations between the lncRNAs and the diseases.

     

DLGCN:基于图卷积网络的药物-lncRNA[]关联预测

为实现高通量识别新的药物-长链非编码RNA(Long non-coding RNA, lncRNA)关联,本文提出了一种基于图卷积网络模型来识别潜在药物-lncRNA关联的方法DLGCN(Drug-LncRNA graph convolution network)。首先,基于药物的结构信息和lncRNA的序列信息分别构建了药物-药物和lncRNA-lncRNA相似性网络,并整合实验证实的药物-lncRNA关联构建了药物-lncRNA异质性网络。然后,将注意力机制和图卷积运算应用于该网络中,学习药物和lncRNA的低维特征,基于整合的低维特征预测新的药物-lncRNA关联。通过效能评估,DLGCN的受试者工作特性曲线下面积(Area under receiver operating characteristic, AUROC)达到0.843 1,优于经典的机器学习方法和常见的深度学习方法。此外,DLGCN预测到姜黄素能够调控lncRNA MALAT1的表达,已被最近的研究证实。DLGCN能够有效预测药物-lncRNA关联,为肿瘤治疗新靶点的识别和抗癌药物的筛选提供了重要参考。     

基于逻辑斯蒂回归模型的鹭科水鸟栖息地适宜性评价

近年来湿地生态系统遭到不同程度破坏,湿地水鸟及其生存空间日益受到威胁。以香港米埔-后海湾湿地为例,收集2003年1月份与鹭科水鸟密切相关的15个自变量和鹭科水鸟实测数据作为因变量构建逻辑斯蒂回归模型,通过筛选获取9个变量因子,分别为土地利用,NDVI,坡度,降雨,TM4纹理,TM3纹理,道路密度,道路距离,人居密度。经Nagelkerke R2检验模型精度达到0.743,拟合度较高。利用模型结果快速聚类,对栖息地进行适宜性分级,分级结果与同期鹭科水鸟实测数据做拟合,精度达到77.4%。最后采集2009年1月份各变量因子数据对回归方程进行时间尺度检验,与同期实测鹭科水鸟数据拟合精度同样达到75.8%,模型具有较好的通用性。     

SCMFMDA: Predicting microRNA-disease associations based on similarity constrained matrix factorization

miRNAs belong to small non-coding RNAs that are related to a number of complicated biological processes. Considerable studies have suggested that miRNAs are closely associated with many human diseases. In this study, we proposed a computational model based on Similarity Constrained Matrix Factorization for miRNA-Disease Association Prediction (SCMFMDA). In order to effectively combine different disease and miRNA similarity data, we applied similarity network fusion algorithm to obtain integrated disease similarity (composed of disease functional similarity, disease semantic similarity and disease Gaussian interaction profile kernel similarity) and integrated miRNA similarity (composed of miRNA functional similarity, miRNA sequence similarity and miRNA Gaussian interaction profile kernel similarity). In addition, the L₂ regularization terms and similarity constraint terms were added to traditional Nonnegative Matrix Factorization algorithm to predict disease-related miRNAs. SCMFMDA achieved AUCs of 0.9675 and 0.9447 based on global Leave-one-out cross validation and five-fold cross validation, respectively. Furthermore, the case studies on two common human diseases were also implemented to demonstrate the prediction accuracy of SCMFMDA. The out of top 50 predicted miRNAs confirmed by experimental reports that indicated SCMFMDA was effective for prediction of relationship between miRNAs and diseases.     

Phenology, seasonal timing and circannual rhythms: towards a unified framework

Phenology refers to the periodic appearance of life-cycle events and currently receives abundant attention as the effects of global change on phenology are so apparent. Phenology as a discipline observes these events and relates their annual variation to variation in climate. But phenology is also studied in other disciplines, each with their own perspective. Evolutionary ecologists study variation in seasonal timing and its fitness consequences, whereas chronobiologists emphasize the periodic nature of life-cycle stages and their underlying timing programmes (e.g. circannual rhythms). The (neuro-) endocrine processes underlying these life-cycle events are studied by physiologists and need to be linked to genes that are explored by molecular geneticists. In order to fully understand variation in phenology, we need to integrate these different perspectives, in particular by combining evolutionary and mechanistic approaches. We use avian research to characterize different perspectives and to highlight integration that has already been achieved. Building on this work, we outline a route towards uniting the different disciplines in a single framework, which may be used to better understand and, more importantly, to forecast climate change impacts on phenology.     

基于图论的景观图表达、分析及应用

景观结构和空间格局一直是景观生态学的核心问题,图论的应用为景观格局的分析提供了一种研究框架,基于图论的景观图逐渐被应用于生物多样性保护的连通性建模和景观规划的决策支持研究,景观图的表达、分析和应用已成为保护生物学和景观生态学研究的热点。本文首先介绍了景观图的图论基础,在Scopus数据库的基础上,检索了1993—2019年在标题、摘要和关键词中出现 “landscape graph”、“connectivity”和“network”词汇的257篇已发表的期刊论文。从年发文量、来源期刊、研究区域、研究机构、景观类型等方面分析了该领域的研究进展和发展趋势。分析发现,2017年之前,发表的期刊论文数量整体呈上升趋势,2017年之后年发文量逐年下降;主要研究力量集中在美国、西班牙、法国、加拿大和中国,发文量占到86.8%。大部分研究成果发表在“Landscape Ecology”、“Landscape and Urban Planning”和“Biological Conservation”期刊上。在研究内容上,景观图表达主要包括点的定义、边的度量和景观的模拟3方面,景观图分析研究包括分析指数、景观图划分两方面。我们重点关注了景观图在生物多样性保护、景观(生态网络)规划和管理、景观影响评价等科学与实践中的应用。基于图论的景观图通过帮助理解景观连通性变化、动物行为和栖息地保护,影响着保护科学和规划实践者。图论对保护科学和规划的影响来自于丰富的理论基础和成熟的研究方法,基于图论的景观图为景观结构和格局的生态学理解提供了跳板,并将继续成为全球研究人员和实践者的重要工具。     

A unified framework for unraveling the functional interaction structure of a biomolecular network based on stimulus-response experimental data

We propose a unified framework for the identification of functional interaction structures of biomolecular networks in a way that leads to a new experimental design procedure. In developing our approach, we have built upon previous work. Thus we begin by pointing out some of the restrictions associated with existing structure identification methods and point out how these restrictions may be eased. In particular, existing methods use specific forms of experimental algebraic equations with which to identify the functional interaction structure of a biomolecular network. In our work, we employ an extended form of these experimental algebraic equations which, while retaining their merits, also overcome some of their disadvantages. Experimental data are required in order to estimate the coefficients of the experimental algebraic equation set associated with the structure identification task. However, experimentalists are rarely provided with guidance on which parameters to perturb, and to what extent, to perturb them. When a model of network dynamics is required then there is also the vexed question of sample rate and sample time selection to be resolved. Supplying some answers to these questions is the main motivation of this paper. The approach is based on stationary and/or temporal data obtained from parameter perturbations, and unifies the previous approaches of Kholodenko et al. (PNAS 99 (2002) 12841-12846) and Sontag et al. (Bioinformatics 20 (2004) 1877-1886). By way of demonstration, we apply our unified approach to a network model which cannot be properly identified by existing methods. Finally, we propose an experiment design methodology, which is not limited by the amount of parameter perturbations, and illustrate its use with an in numero example.     

Animal migrations and parasitism: reciprocal effects within a unified framework

Migrations, i.e. the recurring, roundtrip movement of animals between distant and distinct habitats, occur among diverse metazoan taxa. Although traditionally linked to avoidance of food shortages, predators or harsh abiotic conditions, there is increasing evidence that parasites may have played a role in the evolution of migration. On the one hand, selective pressures from parasites can favour migratory strategies that allow either avoidance of infections or recovery from them. On the other hand, infected animals incur physiological costs that may limit their migratory abilities, affecting their speed, the timing of their departure or arrival, and/or their condition upon reaching their destination. During migration, reduced immunocompetence as well as exposure to different external conditions and parasite infective stages can influence infection dynamics. Here, we first explore whether parasites represent extra costs for their hosts during migration. We then review how infection dynamics and infection risk are affected by host migration, thereby considering parasites as both causes and consequences of migration. We also evaluate the comparative evidence testing the hypothesis that migratory species harbour a richer parasite fauna than their closest free-living relatives, finding general support for the hypothesis. Then we consider the implications of host migratory behaviour for parasite ecology and evolution, which have received much less attention. Parasites of migratory hosts may achieve much greater spatial dispersal than those of non-migratory hosts, expanding their geographical range, and providing more opportunities for host-switching. Exploiting migratory hosts also exerts pressures on the parasite to adapt its phenology and life-cycle duration, including the timing of major developmental, reproduction and transmission events. Natural selection may even favour parasites that manipulate their host's migratory strategy in ways that can enhance parasite transmission. Finally, we propose a simple integrated framework based on eco-evolutionary feedbacks to consider the reciprocal selection pressures acting on migratory hosts and their parasites. Host migratory strategies and parasite traits evolve in tandem, each acting on the other along two-way causal paths and feedback loops. Their likely adjustments to predicted climate change will be understood best from this coevolutionary perspective.     

Bayesian species delimitation combining multiple genes and traits in a unified framework

Delimitation of species based exclusively on genetic data has been advocated despite a critical knowledge gap: how might such approaches fail because they rely on genetic data alone, and would their accuracy be improved by using multiple data types. We provide here the requisite framework for addressing these key questions. Because both phenotypic and molecular data can be analyzed in a common Bayesian framework with our program iBPP, we can compare the accuracy of delimited taxa based on genetic data alone versus when integrated with phenotypic data. We can also evaluate how the integration of phenotypic data might improve species delimitation when divergence occurs with gene flow and/or is selectively driven. These two realities of the speciation process are ignored by currently available genetic approaches. Our model accommodates phenotypic characters that exhibit different degrees of divergence, allowing for both neutral traits and traits under selection. We found a greater accuracy of estimated species boundaries with the integration of phenotypic and genetic data, with a strong beneficial influence of phenotypic data from traits under selection when the speciation process involves gene flow. Our results highlight the benefits of multiple data types, but also draws into question the rationale of species delimitation based exclusively on genetic data.     

Predicting calcium-binding sites in proteins - a graph theory and geometry approach

Identifying calcium-binding sites in proteins is one of the first steps towards predicting and understanding the role of calcium in biological systems for protein structure and function studies. Due to the complexity and irregularity of calcium-binding sites, a fast and accurate method for predicting and identifying calcium-binding protein is needed. Here we report our development of a new fast algorithm (GG) to detect calcium-binding sites. The GG algorithm uses a graph theory algorithm to find oxygen clusters of the protein and a geometric algorithm to identify the center of these clusters. A cluster of four or more oxygen atoms has a high potential for calcium binding. High performance with about 90% site sensitivity and 80% site selectivity has been obtained for three datasets containing a total of 123 proteins. The results suggest that a sphere of a certain size with four or more oxygen atoms on the surface and without other atoms inside is necessary and sufficient for quickly identifying the majority of the calcium-binding sites with high accuracy. Our finding opens a new avenue to visualize and analyze calcium-binding sites in proteins facilitating the prediction of functions from structural genomic information.     

The notch ankyrin domain folds via a discrete, centralized pathway

The Notch ankyrin repeat domain contains seven ankyrin sequence repeats, six of which adopt very similar structures. To determine if folding proceeds along parallel pathways and the order in which repeats become structured during folding, we examined the effect of analogous destabilizing Ala-->Gly substitutions in each repeat on folding kinetics. We find that folding proceeds to an on-pathway kinetic intermediate through a transition state ensemble containing structure in repeats three through five. Repeats two, six, and seven remain largely unstructured in this intermediate, becoming structured in a second kinetic step that leads to the native state. These data suggest that the Notch ankyrin domain folds according to a discrete kinetic pathway despite structural redundancy in the native state and highlight the importance of sequence-specific interactions in controlling pathway selection. This centralized pathway roughly corresponds to a low energy channel through the experimentally determined energy landscape.     

Towards a unified framework for connectivity that disentangles movement and mortality in space and time

Predicting connectivity, or how landscapes alter movement, is essential for understanding the scope for species persistence with environmental change. Although it is well known that movement is risky, connectivity modelling often conflates behavioural responses to the matrix through which animals disperse with mortality risk. We derive new connectivity models using random walk theory, based on the concept of spatial absorbing Markov chains. These models decompose the role of matrix on movement behaviour and mortality risk, can incorporate species distribution to predict the amount of flow, and provide both short‐ and long‐term analytical solutions for multiple connectivity metrics. We validate the framework using data on movement of an insect herbivore in 15 experimental landscapes. Our results demonstrate that disentangling the roles of movement behaviour and mortality risk is fundamental to accurately interpreting landscape connectivity, and that spatial absorbing Markov chains provide a generalisable and powerful framework with which to do so.     

Predicting the start of the birch pollen season at London, Derby and Cardiff, United Kingdom, using a multiple regression model, based on data from 1987 to 1997

Recent changes in weather in North-West Europehave been reflected in the start dates ofpollen seasons. It is therefore necessary toupdate previous models, such as the oneproduced by Jones (1995), so that the modelwill be weighted by current weather patterns. Birch pollen data, collected over a period ofeleven years (1987 to 1997 inclusive) fromthree pollen counting stations in the UK,London, Derby and Cardiff, were analysed todetermine the start dates using the Sum75method. The start dates of the birch pollenseasons of the eleven-year period were thentested for significance against ten-dayaggregated variables of temperature andrainfall for each site. The significantvariables were entered into multiple regressionmodels until the most valid equation for eachsite was found. The models were then tested onthree years not included in their data sets. The models showed mean differences betweenactual and predicted start dates, for theeleven years used, of 1.5, 3 and 5 days atDerby, Cardiff and London respectively. Forthe test years the mean difference was 1, 4.5and 7.5 days at Derby, Cardiff and Londonrespectively. The most powerful model was forDerby where the corresponding meteorologicalstation is at 0.5 km distance and the weakestwas for London where the correspondingmeteorological station is much further away at21 km distance. Weather variables from earlyFebruary to mid March were found to be the mostinfluential on the start dates of the birchpollen season at the three sites.     

Computation of elementary modes: a unifying framework and the new binary approach

Background  

Metabolic pathway analysis has been recognized as a central approach to the structural analysis of metabolic networks. The concept of elementary (flux) modes provides a rigorous formalism to describe and assess pathways and has proven to be valuable for many applications. However, computing elementary modes is a hard computational task. In recent years we assisted in a multiplication of algorithms dedicated to it. We require a summarizing point of view and a continued improvement of the current methods.