本期BMC前沿視角我們有幸邀請到了Genome Biology編委嚴建兵課題組成員為我們解讀最近發表在Genome Biology以及BMC Biology的五篇重要文獻，以饗讀者。
嚴建兵課題組簡介
嚴建兵課題組主要從事玉米基因組和分子育種研究。他們搭建了玉米關聯分析平臺，構建了多個人工分離群體，開發了相應的分析方法，獲得了玉米高質量的參考基因組和變異圖譜，這些研究材料和數據被國內外不同的研究小組廣泛使用。基於這些材料平臺他們課題組探究了玉米關鍵品質和農藝性狀的遺傳學基礎，鑑定了多個關鍵功能基因及其調控網絡，開發了系列功能分子標記，並應用於玉米新品種選育。他課題組另外一個興趣是，植物單細胞測序技術和應用。目前已經開發了玉米四分體、單核和單胚囊等細胞的分離和測序技術，並應用於植物生殖發育，遺傳重組和重編程方面等科學問題的研究。更多信息可以參考：www.maizego.org
嚴建兵教授目前是 Genome Biology, Plant Journal, Science China Life Sciences, Theoretical and Applied Genetics, Journal of Integrative Plant Biology 等期刊的編委會成員。他2015年獲得了國家傑出青年基金的支持，2016年獲得教育部長江學者特聘教授稱號。
本期文章解讀
#1. Title: Leveraging biological and statistical covariates improves the detection power in epigenome-wide association testing
評述人：肖英傑（副教授）
Research
Genome Biology
Date: 2020-04-06
評論：
全表觀組關聯分析 (EWAS) 是研究海量的基因組表觀印記與生物體表型性狀之間關係的一種有效統計手段。假髮現率（FDR）控制是EWAS分析中常用的多重假設檢驗的矯正方法。傳統的FDR方法並不考慮其它協變量的影響，因此統計功效較低。Huang等人利用公共資料庫的61套人類EWAS數據集，測試了5種協變量適應性的FDR方法對改善EWAS統計功效的作用。作者設計並開發了一種Omnibus檢驗，用以評估17個統計學和生物學協變量在數據分析中的信息含量。研究發現，統計學協變量在不同數據集中具有比較通用的信息量，而生物學協變量的信息量則強烈依賴於數據集本身的遺傳結構。進一步分析發現，相比於傳統FDR方法，獨立權重假設法（IHW）和協變量適應性多重檢驗法（CAMT）在不同數據集中均有更好的EWAS統計功效，特別是對於稀疏信號遺傳結構的數據集。作者利用獨立研究鑑定到的衰老和吸菸有關的DMP進行交叉驗證，結果表明協變量適應性FDR方法能更好地檢測到這些DMP。最後，作者提出將不同或相似數據集的EWAS結果作為FDR校正的協變量，同樣能顯著提高EWAS分析的統計功效。總而言之，針對大數據時代的共性難題「多重檢驗」，本研究提出了一種整合統計學和生物學知識的新思路，為人們有效挖掘重要性狀表型背後的生物學機制提供了新的機遇。
Epigenome-wide association study (EWAS) is an effective statistical approach to discover the association between epigenetic marks and biological traits. False discovery rate (FDR) control has been widely used for multiple testing correction. However, traditional FDR control methods do not consider auxiliary covariates, causing reduced statistic power. Huang and his collaborators evaluated the performance of five covariate-adaptive FDR control methods by using 61 public EWAS data. The authors developed an omnibus test to evaluate the informativeness of 17 statistical and biological covariates. They found that statistical covariates are generally more informative and are universal across EWAS datasets, but the informativeness of biological covariates strongly depend on the structure of the datasets. The further analyses revealed that the independent hypothesis weighting (IHW) covariate adaptive multiple testing (CAMT) method are overall more powerful, especially for sparse signals, compared to the traditional ST procedure. Cross validation analysis based on independent data of age- and smoking-associated DMPs revealed that the adaptive FDR control with informative covariates can improve the EWAS power to retrieve biological relevance. In summary, to address the common concern 「multiple testing」 in the era of big data, this research proposed a new idea of integrating statistical and biological knowledge to aid effectively mining the biological mechanisms behind important traits.
#2. Title: Gapless assembly of maize chromosomes using long-read technologies
評述人：吳伸伸 （碩士生）
Short Report
Genome Biology
Date: 2020-05-20
評論：
串聯重複序列的存在給基因組的組裝帶來巨大挑戰，因為其往往會超出目前測序技術的讀長。對於玉米來說，除了富含著絲粒和核糖體DNA（rDNA）序列外，兩大類紐扣重複序列阻礙了玉米基因組的組裝。Liu等人通過將含有Ab10（具有著絲粒狀特性並優先分離於子代）的玉米系與B73回交六代並自交五代創造了一個新的玉米自交系B73-Ab10（BC6F5）。通過同時使用PacBio以及Nanopore的測序技術，並結合基於光學圖譜的合併流程，他們組裝了一個contig N50為162 Mb，並僅僅包含63個contig的玉米基因組B73-Ab10。該高質量的基因組包含無間隙組裝的3號染色體（236 Mb）和9號染色體（162 Mb），以及長53Mb的Ab10減數分裂驅動單倍型。組裝數據還揭示了七個著絲粒和五個異色結節的內部結構，表明主要的串聯重複序列（CentC，knob180和TR-1）不是連續存在的。該研究描述了一種自動合併的組裝方法，該方法可組裝出無間斷的玉米染色體，並顯著提高整個基因組（包括著絲粒和紐結區）的組裝質量。
The most challenging task in genome assembly is to go through the genomic regions containing tandem repeat arrays that exceed the read length of current sequencing technologies. In addition to these arrays that are enriched in centromeres and ribosomal DNA (rDNA), maize contains two abundant classes of knob repeats which largely hamper full genome assembly in Maize. Liu et al., generated a new maize inbred (B73-Ab10) by backcrossing a line containing Ab10, which has centromere-like properties and preferentially segregates to progeny, to the B73 inbred for six times and selfing it for five times (BC6F5). A maize genome (B73-Ab10) composed of 63 contigs with a contig N50 of 162 Mb was assembled by using both PacBio and Nanopore technologies and an optical map-based merging pipeline. This genome includes gapless chromosome 3 (236 Mb) and chromosome 9 (162 Mb), as well as 53 Mb of the Ab10 meiotic driven haplotype. The data also reveals the internal structure of seven centromeres and five heterochromatic knobs, showing that the major tandem repeat arrays (CentC, knob180, and TR-1) are discontinuous. Together, this study describes an automated genome assembly approach that yields gapless maize chromosomes. It dramatically improves contiguity throughout the genome, including centromere and knob regions.
#3. Title: Multiplexed capture of spatial configuration and temporal dynamics of locus-specific 3D chromatin by biotinylated dCas9
評述人：彭勇（博士生）
Method
Genome Biology
Date: 2020-03-05
評論：
真核生物基因組通過三維空間結構介導染色質相互作用調節基因轉錄，繼而控制生物體的生長發育。基因組根據不同解析度可以分為不同結構單元，包括不同的區室，拓撲結構域（TAD）和染色質環。其中，染色質環介導增強子和啟動子的相互作用，以控制組織和發育階段特異性基因的時空表達。目前，解析順式調控元件位點高解析度染色質結構，闡釋順式調控元件作用與基因活性之間關聯仍然具有挑戰性。
應用Hi-C和ChIA-PET等技術已成功構建了全基因組染色質交互圖譜，但受限於解析度，上述技術不能提供時空特異性的染色質相互作用信息。先前，研究人員通過共表達含生物素受體位點的dCas9，細菌BirA生物素連接酶和靶標特異性sgRNA，將結合特定基因座的dCas9複合物在體內進行生物素化，通過基於鏈黴親和素的親和純化以及基於3C的方法分別鑑定作用於特定位點的蛋白質和長距離DNA相互作用。但是，該方法用於對位點特異性染色質相互作用進行無偏分析，由於對細胞數量的要求較高，不適用於原代組織或稀有細胞群體。
本文在之前的基礎上通過使用單轉錄本表達Cas9和BirA、C端生物素化dCas9以及混合sgRNA的方法。在準確引導dCas9結合基因組目標區域的同時，提高了dCas9捕獲的效率。重新設計的系統允許在單個實驗中對幾個到數百個增強子或啟動子的空間構型進行定量分析，從而可以比較基因簇內和簇之間的順式調控元件。通過對紅細胞超級增強子的多元分析，揭示了超級增強子不同層次結構和超級增強子與基因相互作用的獨特模式。針對啟動子區域高通量三維結構的捕獲則鑑定了基因轉錄轉錄和細胞分化過程中增強子-啟動子環的調控功能。該方法為進一步揭示了順式調控元件在基因組中的功能提供了有力工具。
The eukaryotic genome mediates chromatin interaction through a three-dimensional spatial structure. The spatial chromatin structure regulates gene transcription, which in turns controls the growth and development process. The genome can be divided into different structural units based on resolutions, including different compartments, topologically associating domains(TAD) and chromatin loops. The chromatin loop mediates enhancer-promoter interactions to control tissue- and developmental stage-specific gene expression. However, it is still challenging to analyze the high-resolution chromatin structure at certain site and understand the relationship between cis-regulatory elements and gene activity.
Hi-C and ChIA-PET technologies have enabled systematic interrogations of genome-wide landscape of chromatin interactions but they often lack the level of resolution required to evaluate the spatiotemporal organization of locus-specific interactions. Previously, a dCas9-based three-dimensional structure capture method for studying specific sites was developed. By co-expressing dCas9, the biotin receptor site, bacterial BirA biotin ligase and target-specific sgRNA, the dCas9 complex bound locus was biotinylated in vivo and was then separated by affinity purification based on streptavidin. The associated proteins and long-range DNA interactions can be identified. This method was used for unbiased analysis of site-specific chromatin interactions. However, due to the requirements of large number of cells, it is not suitable for primary tissues or rare cell populations.
In this study, by co-expressing BirA and dCas9 from a single transcript and C-terminal biotinylated dCas9 and mixed sgRNAs, the efficiency of dCas9 capture had been largely improved. The new system enables quantitative analysis of the spatial configuration of a few to hundreds of enhancers and promoters in a single experiment. Therefore, cis-regulatory elements within and between gene clusters can be compared. The analysis of erythrocyte super-enhancers reveals the different hierarchies of super-enhancers and the unique pattern of interactions between super-enhancers and genes. The high-throughput capture of the promoter regions identified the determinate function of enhancer-promoter loop in transcription regulation and cell differentiation. This method provides a powerful tool to further study the function of cis-regulatory elements in the genome.
#4. Title: NetConfer: a web application for comparative analysis of multiple biological networks
評述人：桂松濤（博士後）
Software
BMC Biology
Date: 2020-05-19
評論：
複雜的生物系統以及其中各種組分之間的聯繫，往往以基於圖論的網絡形式進行展示。因此，開發可以有效地對基於不同組學數據構建的生物網絡進行比較分析的工具，將有助於更好地挖掘不同組學數據的關聯以及生物網絡系統中的特殊變化。在本研究中，Nagpal團隊開發了NetConfer，一個集成了多種網絡比較分析方法的生物網絡分析網頁應用。基於NetConfer，用戶可以在線對多種生物網絡進行網絡組成相似度評估、網絡關鍵節點的鑑定和比較、網絡最短路徑比較、網絡社區（community）比較以及網絡派系（cliques）比較等工作。為了展示NetConfer的實際應用效果，Nagpal等利用NetConfer對多發性硬化症（multiple sclerosis）腸道微生物相關網絡進行分析，發現腸道微生物組比疾病本身對治療手段更加敏感，可以作為評估幹擾素和克帕松對多發性硬化症治療效果的新指標；Nagpal等還利用NetConfer對致病性和非致病性的結合分枝桿菌（Mycobacterium tuberculosis）感染的人巨噬細胞的時序性基因表達網絡進行了分析，鑑定出了一些可能受到結合分枝桿菌幹擾的基因。最後，為了方便生物學家的使用，NetConfer提供了用戶友好的作業管理和結果展示工具，該應用也提供相應離線版本用於本地大數據的分析。總體而言，NetConfer是一個可以進行多種生物網絡數據的比較、分析和信息挖掘的實用工具。
The complexity of biological systems and the interactions of various components within the system are always presented via graph theory-based networks. Tools enabling comparative analysis of multiple networks help to identify variations across different biological systems. In this study, Nagpal et al. present NetConfer, a web application which implements multiple network comparison and presents them in the form of organized workflows, including assessing similarity of network components, identifying and comparing key nodes, comparing shortest paths, inferring and comparing community structures, as well as comparative analysis of network cliques. In order to demonstrate applications of the NetConfer tool, Nagpal et al. analyzed the gut microbial associated networks in multiple sclerosis, and found that the microbiome is more affected during the treatment as compared to the disease itself, indicating that the microbiome might serve as a valuable mark to evaluate the effectiveness of interferon versus Copaxone. Nagpal et al. also analyzed time series gene networks of human macrophages under differential mycobacterium infection, and identified candidate gene that are affected upon infection. NetConfer tool is developed keeping in mind the requirements of researchers working in the field of biological data analysis with limited programming expertise. A stand-alone version has also been supplemented to accommodate the offline requirement of processing large networks. Collectively, NetConfer is a useful tool for comparative analysis of multiple biological networks.
#5. Title: Accounting for cell type hierarchy inevaluating single cell RNA-seq clustering
評述人：李書彥（碩士生）
Short Report
Genome Biology
Date: 2020-05-25
評論：
相較於傳統的混合RNA-Seq，單細胞RNA-seq可明晰細胞的異質性並解析細胞特異的轉錄和調控，被廣泛應用於疾病、發育和環境應答等研究。細胞分型是單細胞RNA-seq數據分析過程中的重要步驟。所分離的細胞類型是否準確，需要結合通過其他高置信手段獲得的標記基因信息予以確認。目前，評估細胞分型效果主要使用Adjusted Rand Index（ARI）和normalized mutual information（NMI）。但細胞分型問題有其特殊性，真正細胞群體通常具有層級結構，不同的細胞群體之間距離不一致，相關性也不同。傳統評估方式並未考慮層級結構這一因素，因而對軟體分群效果的評估會有偏頗。研究人員在傳統RI和NMI基礎之上，從細胞層次結構中獲得權重，開發出兩種新的矩陣模型weighted Rand index（wRI）和weighted normalized mutual information（wNMI）用於評估單細胞RNA-seq細胞分群結果。在模擬數據測試中，設定有相同的錯誤分群細胞個數，結果1在亞群中將細胞錯誤分類，結果2中將細胞錯誤歸到另一大的亞群。ARI和NMI認為兩種結果分群效果一致，而wRI和wNMI可以區分出兩種結果的差異，並認為結果 1比結果2效果好。利用wRI和wNMI評估monocle，CIDR，Seurat，TSCAN和SC3等多種單細胞RNA-seq細胞分群軟體，發現多種方法的真實性能均高於預期，尤其是CIDR和TSCAN兩款軟體提升較明顯。Seurat和SC3的效果在新的評價方法下相差無幾。這為未來的單細胞數據分析提供了新的工具。
Compared with bulk RNA-seq, scRNA-seq reveals cell–to-cell heterogeneity in transcription, providing unique information to understand biological processes in development, differentiation, disease etiologies and environmental response. To evaluate the performance of a clustering method, the common practice is to compare clustering result with reference markers, which are obtained from another source with high confidence. The most widely used measures to evaluate the performance of a clustering method are the adjusted Rand index (ARI) and the normalized mutual information (NMI). Unlike many other clustering, the true cluster structure for a cell population is often hierarchical, which means that the distance and correlation between different subgroups are not consistent. Traditional methods fail to take this true hierarchy into account in the evaluation of clustering results, leading to assessments that do not accurately reflect the ability to group cells. Researchers modified the traditional RI and NMI methods and developed two new metrics: weighted Rand index (wRI) and weighted normalized mutual information (wNMI), for the evaluation of cell clustering. In a toy example, researchers set the same number of wrong grouped cells to test wRI and wNMI. In result1, the wrong grouped cells are still in the same subgroup but were labeled as wrong, while the wrong grouped cells are classified into different subgroups in result2. Strikingly, ARI and NMI give the two clustering results identical scores. In contrast, both wRI and wNMI could distinguish the two and picked result1. Then researchers apply the new metrics to compare the performances of five popular cell clustering methods, including monocle, CIDR, Seurat, TSCAN and SC3. All of the five software show better performance than reported by RI and NMI, especially in the case of CIDR and TSCAN.The overall performance of Seurat and SC3 is similar. This provides a new tool for future single cell data analysis.
Genome Biologycovers all areas of biology and biomedicine studied from a genomic and post-genomic perspective. Content includes research, new methods and software tools, and reviews, opinions and commentaries. Areas covered include, but are not limited to: sequence analysis; bioinformatics; insights into molecular, cellular and organismal biology; functional genomics; epigenomics; population genomics; proteomics; comparative biology and evolution; systems and network biology; genome editing and engineering; genomics of disease; and clinical genomics. All content is open access immediately on publication.