基于深度卷积神经网络识别可变剪接位点

可变剪接源于多外显子基因生成多个转录本的调控过程。随着高通量测序,尤其是RNA-seq的研究进展,剪接序列和剪接位点可以通过挖掘海量的测序数据进行预测。可变剪接现象拓宽了人们对基因结构和蛋白质亚型的知识。然而现有的短序列比对软件受到随机性比对的影响,产生很多假阳性剪接位点,干扰下游数据分析。本研究发现,可变剪接位点周边序列的结构特征可被深度学习模型提取,并利用深度卷积神经网络识别剪接位点。本研究的模型具有识别率高、计算速度快,模型泛化能力强、鲁棒性高等优势。     

基于miRNA-mRNA联合分析的卵巢癌调控网络构建

研究表明微小RNA(microRNA,miRNA)通过影响转录后基因表达来调节机体功能,并与肿瘤的发生有密切关系。然而目前癌症致病过程的转录调控网络重构大多致力于转录层面的基因表达数据的处理和分析,如何整合转录及转录后不同类型的生物数据以挖掘它们的共调控机制是目前的研究热点之一。基于此,本研究利用联合非负矩阵分解算法融合卵巢癌miRNA数据和基因表达数据形成共模块,其次对特征模块中miRNA的靶基因进行预测分析,最后对mi RNA-mRNA共模块进行转录及转录后共调控网络构建。仿真结果及分子生物学分析表明,通过联合矩阵分析方法所提取的共模块显示出了与卵巢癌致病具有显著的生物相关性和潜在的联系,此外,GO生物过程分析也进一步的揭示了所提取的共模块中miRNA靶基因的生物学功能与卵巢癌致病密切相关。     

利用序列保守模体和局部构象信息预测转录因子结合位点

转录因子结合位点的计算预测是研究基因转录调控的重要环节,但常用的位置特异得分矩阵方法预测特异性偏低.通过深入分析结合位点的生物特征,提出了一种综合利用序列保守模体和局部构象信息的结合位点预测方法,以极大相关得分矩阵作为保守模体的描述模型,并根据二苷参数模型计算位点序列的局部构象,将两类信息得分组合为多维特征向量,在二次判别分析的框架下进行训练和滑动预测.预测过程中还引入了位置信息量以优化似然得分和过滤备选结果.针对大肠杆菌CRP和Fis结合位点数据的留一法测试结果表明,描述模型的改进和多种信息的融合能有效地改善预测方法的性能,大幅度提高特异性.     

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。     

鸡6个功能基因microRNA靶标区域SNP的生物信息学预测

GDF-8、IGF-I、IGF-III、IGF2R、IGFBP2和GHR是鸡的重要经济性状候选基因。利用miRanda和Targetscan软件预测6个基因3′UTR潜在的microRNA靶标, 并发掘靶标区域SNP位点。结果表明: 在6个基因的26个microRNA靶标区域, 共检测到125个SNP位点, 在靶标及其5′和3′邻接等长侧翼区分别检测到47个、44个和35个SNPs位点, 其中12个SNP定位于靶标种子序列互补区。种子序列互补区及其3′侧翼区的SNP位点可能会影响microRNA的调控, 导致家禽的表型变异     

鸡6个功能基因microRNA靶标区域SNP的生物信息学预测

GDF-8、IGF-I,IGF-III、IGF2R、,IGFBP2和GHR是鸡的重要经济性状候选基因.利用miRanda和Targetscan软件预测6个基因3'UTR潜在的microRNA靶标,并发掘靶标区域SNP位点.结果表明:在6个基因的26个microRNA靶标区域,共检测到125个SNP位点,在靶标及其5'和3'邻接等长侧翼区分别检测到47个,44个和35个SNPs位点,其中12个SNP定位于靶标种子序列互补区.种子序列互补区及其3'侧翼区的SNP位点可能会影响microRNA的调控,导致家禽的表型变异.     

MicroRNA对皮肤毛囊发育的调控机制

MicroRNA是参与转录后水平表达调控的重要因子, 在病理上成为药物作用的潜在靶点, 在生理上成为表型调控的潜在位点。目前, 对于microRNA的功能已有一定了解, 但其在皮肤毛囊发育中的作用机制还不完全清楚。近年来, 高通量测序技术为microRNA的鉴定提供了更准确、快速的途径, 研究发现一些microRNA能够影响皮肤毛囊细胞的分化和增殖, 其相关靶基因在调控毛囊周期性生长的过程中充当重要角色。文章综述了近年来microRNA在皮肤毛囊生长发育调控机制研究领域所取得的成果, 以期为后续开展绒山羊毛囊生长相关microRNA作用机制研究提供借鉴。     

小鼠肝脏MicroRNA的提取与分离

本文对小鼠肝脏microRNA进行分离制备及相关性功能研究.采用poly(A)加尾方法,分离提取小鼠肝脏中的microRNA;利用T4 RNA连接酶,在microRNA两端加上连接引物,通过RT PCR制备microRNA的cDNA文库;经过对cDNA文库的质粒连接,克隆测序获得microRNA. 结果显示：获得4条有效序列,其中包括2条已知序列miR 122和2条未知小分子RNA序列.经查证, 2条microRNA片段在miRBase库中未有记录,在二级结构分析中可形成茎环结构,符合microRNA的特征. BLAST比对显示： 2个序列位于小鼠载脂蛋白B基因和小鼠28S核糖体RNA上,可能与基因调控有关,其功能有待进一步研究.     

基于位置权重矩阵的核小体识别及功能分析

为研究高通量的人类CD4+T细胞的核小体定位模式,使用迭代算法对核小体定位模式进行分类,并利用位置权重矩阵方法分别构建稳定核小体定位序列、动态核小体定位序列和连接区序列模型,通过十倍交叉验证评估模型性能,并与Segal方法与弯曲度方法进行比较,发现位置权重矩阵方法在敏感性、精度和准确性方面都具有一定优越性。同时采用滑窗法在全基因组选取候选序列进行核小体识别,挖掘核小体定位相关基因,并进行基因生物学进程功能富集分析,发现稳定与动态核小体、真实与潜在核小体对应的基因所参与调控的生物学过程各有不同,但也有一些生物学过程为不同类别核小体所共有,例如对细胞内大分子的调控功能。     

植物microRNA及其抗胁迫作用机制研究进展

植物中广泛存在的microRNA是一类长度约20~24 nt的非编码RNA,它作为负调控因子,通过降解目的基因或抑制目的基因的翻译作用,在转录后水平调控基因的表达.植物microRNA参与生长发育等功能的调控,并在抗生物或非生物胁迫中发挥着重要的作用,如调控植物体内磷、硫的代谢平衡及应对氧化胁迫等生理过程.本文对植物microRNA的特点、形成、作用机制、功能及研究技术方法进行了综述.     

k-gram方法识别microRNA前体

MicroRNAs(miRNAs)是动植物中较短的参与调控基因表达的功能性非编码RNA序列.第一个miRNA是通过实验手段发现的,然而通过实验手段识别miRNA在技术上仍然具有很大的挑战性和不完整性.因此,miRNA基因识别需要寻求计算方法来弥补实验方法的不足.提出了一个全新的miRNA前体的识别方法.在构造识别模型中,把初级序列和序列二级结构相结合,采用k-gram方法把序列信息映射到高维特征空间中,然后通过特征选取方法提取特征,并用这些特征为miRNA前体的识别构造了基于SVM的识别模型.同时,采用隐马尔可夫模型(HMM)的学习方法进行了比较.实验结果表明,该方法是有效的,可以达到较高的敏感性和特异性.     

k-gram方法识别microRNA前体

MicroRNAs (miRNAs) 是动植物中较短的参与调控基因表达的功能性非编码RNA序列. 第一个miRNA是通过实验手段发现的，然而通过实验手段识别miRNA在技术上仍然具有很大的挑战性和不完整性. 因此，miRNA基因识别需要寻求计算方法来弥补实验方法的不足. 提出了一个全新的miRNA前体的识别方法. 在构造识别模型中，把初级序列和序列二级结构相结合，采用k-gram方法把序列信息映射到高维特征空间中，然后通过特征选取方法提取特征，并用这些特征为miRNA前体的识别构造了基于SVM的识别模型. 同时，采用隐马尔可夫模型(HMM)的学习方法进行了比较. 实验结果表明，该方法是有效的，可以达到较高的敏感性和特异性.     

Integrating node embeddings and biological annotations for genes to predict disease-gene associations

Background

Predicting disease causative genes (or simply, disease genes) has played critical roles in understanding the genetic basis of human diseases and further providing disease treatment guidelines. While various computational methods have been proposed for disease gene prediction, with the recent increasing availability of biological information for genes, it is highly motivated to leverage these valuable data sources and extract useful information for accurately predicting disease genes.

Results

We present an integrative framework called N2VKO to predict disease genes. Firstly, we learn the node embeddings from protein-protein interaction (PPI) network for genes by adapting the well-known representation learning method node2vec. Secondly, we combine the learned node embeddings with various biological annotations as rich feature representation for genes, and subsequently build binary classification models for disease gene prediction. Finally, as the data for disease gene prediction is usually imbalanced (i.e. the number of the causative genes for a specific disease is much less than that of its non-causative genes), we further address this serious data imbalance issue by applying oversampling techniques for imbalance data correction to improve the prediction performance. Comprehensive experiments demonstrate that our proposed N2VKO significantly outperforms four state-of-the-art methods for disease gene prediction across seven diseases.

Conclusions

In this study, we show that node embeddings learned from PPI networks work well for disease gene prediction, while integrating node embeddings with other biological annotations further improves the performance of classification models. Moreover, oversampling techniques for imbalance correction further enhances the prediction performance. In addition, the literature search of predicted disease genes also shows the effectiveness of our proposed N2VKO framework for disease gene prediction.

     

Explainable deep transfer learning model for disease risk prediction using high-dimensional genomic data

Building an accurate disease risk prediction model is an essential step in the modern quest for precision medicine. While high-dimensional genomic data provides valuable data resources for the investigations of disease risk, their huge amount of noise and complex relationships between predictors and outcomes have brought tremendous analytical challenges. Deep learning model is the state-of-the-art methods for many prediction tasks, and it is a promising framework for the analysis of genomic data. However, deep learning models generally suffer from the curse of dimensionality and the lack of biological interpretability, both of which have greatly limited their applications. In this work, we have developed a deep neural network (DNN) based prediction modeling framework. We first proposed a group-wise feature importance score for feature selection, where genes harboring genetic variants with both linear and non-linear effects are efficiently detected. We then designed an explainable transfer-learning based DNN method, which can directly incorporate information from feature selection and accurately capture complex predictive effects. The proposed DNN-framework is biologically interpretable, as it is built based on the selected predictive genes. It is also computationally efficient and can be applied to genome-wide data. Through extensive simulations and real data analyses, we have demonstrated that our proposed method can not only efficiently detect predictive features, but also accurately predict disease risk, as compared to many existing methods.     

Refining microRNA target predictions: Sorting the wheat from the chaff

microRNAs are short RNAs that reduce gene expression by binding to their targets. The accurate prediction of microRNA targets is essential to understanding the function of microRNAs. Computational predictions indicate that all human genes may be regulated by microRNAs, with each microRNA possibly targeting thousands of genes. Here we discuss computational methods for identifying mammalian microRNA targets and refining them for further experimental validation. We describe microRNA target prediction resources and procedures and how they integrate with various types of experimental techniques that aim to validate them or further explore their function. We also provide a list of target prediction databases and explain how these are curated.     

Cepred: Predicting the Co-Expression Patterns of the Human Intronic microRNAs with Their Host Genes

Identifying the tissues in which a microRNA is expressed could enhance the understanding of the functions, the biological processes, and the diseases associated with that microRNA. However, the mechanisms of microRNA biogenesis and expression remain largely unclear and the identification of the tissues in which a microRNA is expressed is limited. Here, we present a machine learning based approach to predict whether an intronic microRNA show high co-expression with its host gene, by doing so, we could infer the tissues in which a microRNA is high expressed through the expression profile of its host gene. Our approach is able to achieve an accuracy of 79% in the leave-one-out cross validation and 95% on an independent testing dataset. We further estimated our method through comparing the predicted tissue specific microRNAs and the tissue specific microRNAs identified by biological experiments. This study presented a valuable tool to predict the co-expression patterns between human intronic microRNAs and their host genes, which would also help to understand the microRNA expression and regulation mechanisms. Finally, this framework can be easily extended to other species.     

microRNA target predictions across seven Drosophila species and comparison to mammalian targets

microRNAs are small noncoding genes that regulate the protein production of genes by binding to partially complementary sites in the mRNAs of targeted genes. Here, using our algorithm PicTar, we exploit cross-species comparisons to predict, on average, 54 targeted genes per microRNA above noise in Drosophila melanogaster. Analysis of the functional annotation of target genes furthermore suggests specific biological functions for many microRNAs. We also predict combinatorial targets for clustered microRNAs and find that some clustered microRNAs are likely to coordinately regulate target genes. Furthermore, we compare microRNA regulation between insects and vertebrates. We find that the widespread extent of gene regulation by microRNAs is comparable between flies and mammals but that certain microRNAs may function in clade-specific modes of gene regulation. One of these microRNAs (miR-210) is predicted to contribute to the regulation of fly oogenesis. We also list specific regulatory relationships that appear to be conserved between flies and mammals. Our findings provide the most extensive microRNA target predictions in Drosophila to date, suggest specific functional roles for most microRNAs, indicate the existence of coordinate gene regulation executed by clustered microRNAs, and shed light on the evolution of microRNA function across large evolutionary distances. All predictions are freely accessible at our searchable Web site http://pictar.bio.nyu.edu.     

Protein Function Prediction: From Traditional Classifier to Deep Learning

Deep learning demonstrates greater competence over traditional machine learning techniques for many tasks. In last several years, deep learning has been applied to protein function prediction and a series of good achievements has been obtained. These findings extensively advanced our understanding of protein function. However, the accuracy of protein function prediction based upon deep learning still has yet to be improved. In article number 1900019, Issue 12, Zhang et al. construct DeepFunc, a deep learning framework using derived feature information of protein sequence and protein interactions network. They find that implementing DeepFunc for protein function prediction is more accurate than using DeepGO, a similar method reported previously. Meanwhile, they find that the method of combining multiple derived feature information in DeepFunc is much better than the method of using only single derived feature information. Due to its fully exploiting feature representation learning ability, deep learning with more derived feature information will enable it to be a promising method for solving more complicated protein function prediction problems and other bioinformatics challenges. Recent researches have provided some major insights into the value for using deep learning to protein function prediction problem.