僅供學(xué)術(shù)交流，嚴(yán)禁轉(zhuǎn)載及商用圃泡！原內(nèi)容來(lái)自于10X Genomics官方支持文檔。

什么是Loupe Cell Browser? - What is Loupe

簡(jiǎn)介 - Introduction

Loupe Cell Browser is a desktop application for Windows and MacOS that allows you to quickly and easily visualize and analyze 10x Chromium? Single Cell ATAC data. It is optimized for finding significant peaks and distinguishing transcription factor motifs, identifying cell types, comparing chromatin accessibility between groups of cells, and exploring substructure within cell clusters. Loupe is named for a jeweler's loupe, which is used to inspect gems.

Loupe Cell Browser是一款能夠在Windows和MacOS上運(yùn)行的桌面應(yīng)用，可以讓你快速触创、方便的可視化和分析10x Chromium? Single Cell ATAC數(shù)據(jù)。該軟件用于尋找顯著peaks为牍，區(qū)分轉(zhuǎn)錄因子motif哼绑，識(shí)別細(xì)胞類型，比較細(xì)胞群間染色質(zhì)可接近性碉咆，以及探索細(xì)胞群內(nèi)的亞群抖韩。Loupe是根據(jù)珠寶商用來(lái)檢查寶石的小型放大鏡(Loupe)來(lái)命名的。

overview_atac.png

Loupe Cell Browser opens .cloupe files generated by the Cell Ranger ATAC pipeline. Once your data is in Loupe Cell Browser, you can rapidly explore and gain insights from the data without writing a line of code.

Loupe Cell Browser的輸入文件為Cell Ranger ATAC流程生成的 .cloupe 文件疫铜。當(dāng)你把數(shù)據(jù)加載到Loupe Cell Browser中時(shí)茂浮，你就可以快速的探索數(shù)據(jù)，從中獲得更多的見(jiàn)解壳咕，而不用寫一行代碼席揽。

Loupe Cell Browser is built to accelerate the following applications:

Loupe Cell Browser為了加速以下應(yīng)用而構(gòu)建：

• Identifying Cell Types

Find promoters and transcription factor motifs that differentiate distinct cell types and functional groups using Accessibility mode.

• 識(shí)別細(xì)胞類型

利用可接近性模式發(fā)現(xiàn)區(qū)別特異細(xì)胞類型及功能群組的啟動(dòng)子和轉(zhuǎn)錄因子motif。

• Analyzing Differential Accessibility

Use the ATAC Peak Viewer to pinpoint the location of differentially accessible genomic regions with putative regulatory function.

• 分析差異可接近性

使用ATAC Peak Viewer精確查找差異可接近基因組區(qū)域的位置谓厘，推斷其調(diào)控功能幌羞。

• Finding Significant Features

Create custom clusters and use differential accessibility tools to identify precise cell groups within your data.

• 發(fā)現(xiàn)顯著性特征

創(chuàng)建自定義分群并利用差異可接近性工具來(lái)識(shí)別數(shù)據(jù)集中精確細(xì)胞群組。

• Sharing Results

Save features of interest, export data tables, and capture screenshots of your single-cell ATAC data.

• 分享結(jié)果

保存感興趣的特征竟稳，導(dǎo)出數(shù)據(jù)表属桦，保存single-cell ATAC數(shù)據(jù)的截圖熊痴。

To walk through the features and uses of Loupe Cell Browser for single-cell ATAC data, start the ATAC tutorial.

為了更快的了解針對(duì)single-cell ATAC數(shù)據(jù)分析的Loupe Cell Browser的特性及使用方法，從ATAC教程開始吧聂宾！

Loupe Cell Browser can also be used to analyze Chromium? Single Cell 5′ and 3′ gene expression and immune profiling data. To learn how, consult the Loupe Cell Browser for Gene Expression page.

Loupe Cell Browser也可以用來(lái)分析Chromium? Single Cell 5′ 和 3′基因表達(dá)以及免疫分析數(shù)據(jù)愁拭。想了解更多？移步Loupe Cell Browser for Gene Expression頁(yè)面

Finally, if you have already used Loupe Cell Browser to investigate gene expression, and are new to ATAC, you should read Differences from Gene Expression.

最后亏吝，如果你已經(jīng)用Loupe Cell Browser分析過(guò)基因表達(dá)數(shù)據(jù)岭埠，但是剛了解ATAC，你應(yīng)該讀一讀同“基因表達(dá)差異”小節(jié)蔚鸥。

下載

https://support.10xgenomics.com/single-cell-atac/software/downloads/latest

安裝：Loupe Cell Browser

下載和安裝

首先惜论，從下載頁(yè)下載Loupe Cell Browser。

windows

Loupe Cell Browser for Windows is distributed as a self-installing .exe file. Double-click on the downloaded file to install.

Loupe Cell Browser for Windows作為一個(gè)自安裝的.exe文件進(jìn)行分發(fā)止喷。雙擊下載好的文件開始安裝馆类。

win.png

You will then be prompted to choose an installation folder. After installation, you will be able to open Loupe Cell Browser by double-clicking on the desktop icon, or double-clicking on a .cloupe file in your file system.

接著會(huì)提示你選擇要安裝到的文件夾。安裝結(jié)束后弹谁，你就可以通過(guò)雙擊桌面圖標(biāo)乾巧，或者雙擊一個(gè).cloupe文件來(lái)打開Loupe Cell Browser。

macOS

Loupe Cell Browser for the Mac is distributed as a DMG file. Open this file by double-clicking on it. Then install Loupe by dragging the Loupe icon into the Applications folder. You can then start Loupe by double-clicking on its app icon.

Loupe Cell Browser for the Mac作為一個(gè)DMG文件進(jìn)行分發(fā)预愤。雙擊打開這個(gè)文件沟于。接著拖動(dòng)Loupe圖標(biāo)到Applications文件夾來(lái)開始安裝。隨后通過(guò)雙擊相應(yīng)應(yīng)用圖標(biāo)來(lái)啟動(dòng)Loupe植康。

mac-3.png

You will then be able to open Loupe Cell Browser within the Applications folder, or by double-clicking on a Loupe Cell Browser .cloupe file in your file system.

你可以在Application文件夾中旷太，或雙擊一個(gè)Loupe Cell Browser .cloupe文件來(lái)打開Loupe Cell Browser。

與基因表達(dá)的區(qū)別 - Differences from Gene Expression

If you've used Loupe Cell Browser before to analyze gene expression, you will find exploring ATAC data familiar in some ways, and different in others. The Cell Ranger ATAC algorithm documentation covers algorithms and analysis in more detail, but in short, here are some key things to keep in mind when looking at ATAC data:

如果你之前使用過(guò)Loupe Cell Browser來(lái)分析基因表達(dá)销睁，你會(huì)發(fā)現(xiàn)在探索ATAC數(shù)據(jù)時(shí)供璧，一些方式上兩者相似，其他方面則略有不同冻记。Cell Ranger ATAC算法文檔包含了算法和分析方面的更多細(xì)節(jié)睡毒，但是簡(jiǎn)短來(lái)講，以下是在查看ATAC數(shù)據(jù)時(shí)一些關(guān)鍵性的東西：

• UMI count per cell is the unit of gene expression. Cut sites per cell is the unit of accessibility.

• UMI count per cell是基因表達(dá)的單位冗栗。Cut sites per cell是可接近性的單位演顾。

• Genes are the rows of a gene expression matrix. Peaks are the rows of a chromatin accessibility matrix.

• 基因表達(dá)矩陣中每一行是基因。染色質(zhì)可接近性矩陣中每一行是peak贞瞒。

• Peaks are genomic regions where there were significant upticks in fragment cut sites, which indicate regions of open chromatin. They are named by their location (e.g., "chr1:10244-10510")

• Peaks是基因組區(qū)域偶房，這些區(qū)域在片段切割位點(diǎn)(fragment cut sites)具有顯著提升，表明為開放染色質(zhì)區(qū)域军浆。它們通過(guò)其位置來(lái)命名（例如“Chr1:10244-10510”）棕洋。

• Unlike genes, peaks are likely to be different between different datasets.

• 不同于基因，peaks在不同數(shù)據(jù)集間更傾向于不同乒融。

• There are typically more distinct peaks in an ATAC dataset than there are genes in a reference.

• 通常掰盘，ATAC數(shù)據(jù)集中的峰數(shù)要比參考基因組中的基因多

• The dynamic range of gene expression per cell is typically much wider than the dynamic range of cut sites per peak per cell. This means that you will often use aggregate features (see below) to separate data.

• Gene expression per cell的動(dòng)態(tài)范圍相比cut sites per peak per cell的動(dòng)態(tài)范圍要更寬摄悯。這意味著你需要經(jīng)常使用累加特性（見(jiàn)下文）來(lái)分離數(shù)據(jù)。

• In addition to peaks, there are several aggregate feature types which can be also used to differentiate cells:

• 除了peaks之外愧捕，還有一些其他累加特性類型可用來(lái)進(jìn)行細(xì)胞區(qū)分：

• Promoter sums, which are the sums of cut sites per cell (within peaks) which are close to one of the transcription start sites for that gene. These features are named "(Gene) Sum". Not all peaks are associated with a gene.

• 啟動(dòng)子總和奢驯，是接近該基因的轉(zhuǎn)錄起始位點(diǎn)之一的cut sites per cell (within peaks)的總和。這些特征被命名為"(Gene)Sum"次绘。并非所有peaks都被關(guān)聯(lián)到了一個(gè)基因瘪阁。

• Transcription factor motifs, which are the sums of cut sites per cell which fall within peaks associated with a motif by the Cell Ranger ATAC pipeline. Motif features are named after the motifs themselves (e.g., "SPI1"). A peak is usually associated with multiple motifs.

• 轉(zhuǎn)錄因子motif，是位于被Cell Ranger ATAC流程關(guān)聯(lián)了motif的peak中的cut sites per cell的總和邮偎。Motif特征是以motif本身命名的（例如“SPI1”）管跺。一個(gè)peak通常會(huì)關(guān)聯(lián)多個(gè)motifs。

• An ATAC dataset will take up several times as much disk space (per cell) than a gene expression dataset.

• 一個(gè)ATAC數(shù)據(jù)集會(huì)占用基因表達(dá)數(shù)據(jù)集幾倍的磁盤空間禾进。

• To see fragment locations per cluster in high resolution, you will need access to the fragments.tsv.gz file for that run, generated by the Cell Ranger ATAC pipeline. These files are typically several times larger than the .cloupe file, which is why they are not bundled. You can either specify the location of this file on a locally mounted file system, or on the web via a URL.

在高分辨率下觀察每個(gè)分群片段位置豁跑，需要適用到當(dāng)次Cell Ranger ATAC流程分析生成 fragments.tsv.gz 的文件。這些文件通常比 .cloupe 文件大幾倍泻云，這也是為什么它沒(méi)有被捆綁在其中的原因艇拍。你可以指定該文件在本地掛載的文件系統(tǒng)中的位置，或通過(guò)URL訪問(wèn)web上的位置宠纯。

Loupe Cell Browser ATAC 教程

Welcome to the Loupe Cell Browser ATAC tutorial. In the next few pages, we will be finding significant features, analyzing differential accessibility patterns, and exploring substructure within a real-world dataset. Along the way, we'll touch on most of the features of Loupe Cell Browser.

歡迎來(lái)到Loupe Cell Browser ATAC教程部分卸夕。在接下來(lái)的幾頁(yè)中，我們將尋找顯著特征征椒，分析差異可接近性模式娇哆，以及探索真實(shí)數(shù)據(jù)中的亞群。通過(guò)學(xué)習(xí)本教程勃救，我們將接觸到Loupe Cell Browser的大部分功能。

環(huán)境建立 - Setup

Before beginning the tutorial, make sure you have downloaded Loupe Cell Browser. If this is your first time working with Loupe Cell Browser, you can access the ATAC tutorial dataset by clicking on the "ATACTutorial.cloupe" link on the Recent Files page:

在開始學(xué)習(xí)教程之前治力，請(qǐng)確定你已經(jīng)下載并安裝了Loupe Cell Browser蒙秒。如果這是你第一次使用Loupe Cell Browser，你可以通過(guò)點(diǎn)擊“近期文件”頁(yè)面中的“ATACTutorial.cloupe”來(lái)獲取ATAC教程數(shù)據(jù)集宵统。

first_load_atac.png

You can also access the tutorial dataset by clicking on 'Load ATAC Tutorial Dataset' from the Help menu.

你也可以通過(guò)點(diǎn)擊幫助菜單中的“Load ATAC Tutorial Dataset”來(lái)獲得教程數(shù)據(jù)集晕讲。

About the ATAC Tutorial Dataset

The ATAC tutorial dataset contains the results of a cellranger-atac run over a set of human peripheral blood mononuclear cells, with the standard Chromium? Single-Cell ATAC protocol. The targeted cell count was 5,000; the observed barcode count from the pipeline was 5,335.

ATAC教程數(shù)據(jù)集包含了一個(gè)使用cellranger-atac分析人外周血單核細(xì)胞（根據(jù)標(biāo)準(zhǔn)Chromium? Single-Cell ATAC protocol）的結(jié)果。靶細(xì)胞計(jì)數(shù)為5000马澈，流程分析觀測(cè)到的barcode計(jì)數(shù)為5335瓢省。

Now that you've loaded the file and familiarized yourself with some ATAC data concepts, click here to tour the user interface.

現(xiàn)在你已經(jīng)加載了文件，了解了一些ATAC數(shù)據(jù)的概念痊班，可以開始學(xué)習(xí)用戶界面了勤婚。

Loupe Cell Browser: 用戶界面

With the ATACTutorial dataset loaded, let's take a quick tour of the Loupe Cell Browser user interface for ATAC datasets.

加載好ATACTutorial數(shù)據(jù)集，我們來(lái)快速瀏覽一下Loupe Cell Browser的用戶界面涤伐。

Barcode Plot

The workspace is centered around the barcode plot, in which single points representing cell barcodes are shown in a variety of projections. Each point represents a single barcode, the vast majority of which represent a single cell. The default projection is the t-SNE plot created by the Cell Ranger ATAC pipeline. Cell Ranger ATAC generates this plot by identifying the most significant peak vectors using dimensionality reduction techniques, and then processing the lower-dimension matrix through t-SNE to produce a two-dimensional scatter plot. You may also view a projection that plots cut site counts in a peak, near a gene promoter, or within a certain motif on two-dimensional axes. The selector at the upper right of the barcode plot allows you to toggle between projections.

整個(gè)工作空間以barcode圖為中心馒胆，其中代表細(xì)胞barcode的單個(gè)點(diǎn)在各種投射中展示缨称。每一個(gè)點(diǎn)代編一個(gè)單一barcode，絕大多數(shù)代表一個(gè)單一細(xì)胞祝迂。默認(rèn)的投射是Cell Ranger ATAC流程生成的t-SNE圖睦尽。Cell Ranger ATAC通過(guò)使用降維手段識(shí)別最顯著的peak向量，接著使用t-SNE來(lái)處理低維矩陣以生成二維散點(diǎn)圖型雳。你也可以在一個(gè)二維圖中查看peak中当凡，臨近基因啟動(dòng)子，或在特定motif中的切割位點(diǎn)計(jì)數(shù)投射纠俭。

navigation_atac.png

You can click-and-drag the mouse over the cells to reposition the plot, and use the mouse wheel or track pad to smoothly zoom in and out. You'll see cluster labels as you move your mouse over the plot, which is useful for data that has a high number of precomputed clusters. Cells are colored by the active legend in the sidebar.

你可以在圖上點(diǎn)擊并拖動(dòng)鼠標(biāo)來(lái)改變展示位置沿量，使用鼠標(biāo)滾輪或觸摸板平滑的放大或縮小。當(dāng)你鼠標(biāo)在圖上滑動(dòng)時(shí)會(huì)看到分群標(biāo)識(shí)柑晒，這對(duì)有大量預(yù)計(jì)算的分群的數(shù)據(jù)很有幫助欧瘪。細(xì)胞根據(jù)側(cè)邊欄中活動(dòng)圖例進(jìn)行著色。

工具盒 - The Toolbox

On the left side of the window is the toolbox. When you move your mouse over the toolbox buttons, you will see an explanation of what each button does. Use the toolbox to open files, save your work, control zoom, select cells for manual categorization, export screenshots of the current plot view, and split the barcode plot into individual clusters. Clicking the 10x button will return you to the home screen and Recent Files list.

窗口左側(cè)是工具盒匙赞。當(dāng)你滑動(dòng)鼠標(biāo)到工具盒相應(yīng)按鈕上佛掖，你會(huì)看到關(guān)于這個(gè)按鈕的功能解釋。使用工具盒中的工具涌庭，你可以打開文件芥被，保存工作，控制放大縮小坐榆，選擇細(xì)胞進(jìn)行手動(dòng)分類拴魄，導(dǎo)出當(dāng)前展示圖的截屏，將barcode圖按分群分隔席镀。點(diǎn)擊10x按鈕會(huì)返回起始頁(yè)和近期文件列表匹中。

模式選擇器&側(cè)邊欄 - Mode Selector & Sidebar

Use the mode selector at the top right corner of the workspace to switch between Loupe Cell Browser's different modes. There are two modes in Loupe Cell Browser 3.0 for ATAC data: Categories mode, where you can see the different cluster assignments for all the cells, and Accessibility mode, where you can overlay quantitative cut site count information atop the barcodes. Switching between modes will apply mode-specific coloring to the graph, and change the sidebar to reveal mode-specific functionality.

使用工作區(qū)右上角的模式選擇器來(lái)切換Loupe Cell Browser的不同模式。對(duì)于ATAC數(shù)據(jù)豪诲，Loupe Cell Browser 3.0有兩種模式：分類模式顶捷，在此模式下你可以看到對(duì)全部細(xì)胞進(jìn)行的不同分群；可接近性模式屎篱，在此模式下你可以在badrcodes之上疊加定量切割位點(diǎn)計(jì)數(shù)信息服赎。切換不同的模式會(huì)對(duì)圖應(yīng)用模式特異著色，且改變側(cè)邊欄來(lái)展示模式特有功能。

分類模式 - Categories Mode

Cell Ranger ATAC pipelines compute and produce clusterings from two algorithms: a graph-based clustering algorithm and by either K-Means or K-Medoids clustering. A selector at the top of the Categories sidebar allows you to switch between these clusterings, or other clusterings that you can create yourself within Loupe Cell Browser.

Cell Ranger ATAC流程通過(guò)兩種算法計(jì)算和生成聚類：基于圖的聚類算法，K-Means或K-Medoids魏滚。分類側(cè)邊欄頂部的選擇器可以讓你在這些聚類分群中切換蒋伦，或在Loupe Cell Browser中創(chuàng)建其他自定義分群。

show_hide_atac.png

You may hide, show and highlight individual clusters within a category by using the sidebar. To highlight a cluster, click on the cluster name within the legend. To toggle the appearance of a cluster, click on the checkbox next to the cluster name. Finally, you may hide or show all clusters within a category by clicking on the menu with three dots, to the right of the category selector.

利用側(cè)邊欄，你也可以對(duì)一個(gè)分類中的單個(gè)分群進(jìn)行隱藏，展示砚殿，高亮芽死。想要高亮一個(gè)分群乏梁，單擊圖例中相應(yīng)的分群名稱。決定是否展示一個(gè)分群关贵，單擊分群名稱邊上的復(fù)選框遇骑。最后，你也可以通過(guò)點(diǎn)擊分類選擇器右邊三個(gè)點(diǎn)的菜單來(lái)隱藏或展示一個(gè)分類中的所有分群揖曾。

You also may also rename and recolor clusters by right-clicking on a cluster name or color, and selecting the desired option from the pop-up menu.

你也可以通過(guò)右鍵點(diǎn)擊分群名稱落萎，從彈出的菜單中選擇期望的選項(xiàng)對(duì)分群進(jìn)行重命名或重新著色。

可接近性模式 - Accessibility Mode

accessibility.png

In Accessibility mode, you see a graphical representation of chromatin accessibility across your dataset. You can view the number of cut sites detected per cell within individual peaks, near the promoter regions of particular genes, or in total across the entire genome, through the Peak Sum feature. You may also view Z-scores of transcription factor motif counts per barcode, look at one or more features at a time, load and save lists of features for analyzing across multiple datasets, and look at the density of cut site counts across your data. We will explore Accessibility mode more in-depth when looking for cell types.

在可接近性模式下炭剪，你可以看到一個(gè)整個(gè)數(shù)據(jù)集染色質(zhì)可接近性的圖像化展示练链。你可以查看Peak Sum特性查看單個(gè)peak內(nèi)、特定基因啟動(dòng)子區(qū)域附近或整個(gè)基因組范圍內(nèi)每個(gè)細(xì)胞檢測(cè)到的切割位點(diǎn)的數(shù)量奴拦。還可以查看每個(gè)barcode的轉(zhuǎn)錄因子motif計(jì)數(shù)的Z-score值媒鼓，一次查看一個(gè)或多個(gè)特性，加載和保存用于跨多個(gè)數(shù)據(jù)集分析的特性列表错妖，并查看整個(gè)數(shù)據(jù)的切割位點(diǎn)計(jì)數(shù)的密度绿鸣。在尋找細(xì)胞類型時(shí)，我們將更深入地探討可接近性模式暂氯。

特征表&Peak查看器 - Feature Table & Peak Viewer

peak_viewer.png

The panel on the bottom of the workspace does double duty for ATAC data in Loupe Cell Browser. The Feature Table shows information about differentially accessible peaks, transcription factor motifs, or promoter sums between clusters. The Peak Viewer shows the differential distribution of peaks and cut sites per cluster within the genome. You can use the toolbar at bottom left to toggle between the two panels. When you first load the ATAC dataset, you will see the feature table, preloaded with the transcription factor motifs that are most significantly different between the clusters in the active category. Selecting the Peak Viewer will by default show the first five peaks in the genome, and their distribution within the active set of clusters. We will explore the Feature Table in depth in Significant Features, and cover how to hone in on regions of interest on the Peak Viewer page.

Loupe Cell Browser工作區(qū)下方的面板在分析ATAC數(shù)據(jù)時(shí)具有雙重功能潮模。特征表展示了關(guān)于差異可接近性peaks，轉(zhuǎn)錄因子motifs痴施，分群之間啟動(dòng)子總和的相關(guān)信息擎厢。Peak Viewer展示了基因組中peaks的差異分布以及cut sites per cluster的差異分布。你可以使用左下方的工具條來(lái)切換兩個(gè)面板辣吃。當(dāng)你第一次載入ATAC數(shù)據(jù)集時(shí)动遭，你將看到feature表，其中預(yù)加載了在活動(dòng)類別中的分群之間差異最為顯著的轉(zhuǎn)錄因子motifs神得。選擇Peak Viewer初始后默認(rèn)會(huì)顯示基因組中前五個(gè)peaks沽损，以及這些peaks在活動(dòng)分群集中的分布。我們將在顯著特征一節(jié)進(jìn)一步深入探索特征表循头，介紹如何在Peak Viewer頁(yè)面上找到感興趣的區(qū)域。

Now that you are familiar with the user interface, let's explore the data.

現(xiàn)在你已經(jīng)熟悉了用戶界面炎疆，我們開始探索數(shù)據(jù)吧卡骂。

細(xì)胞類型識(shí)別 - Identifying Cell Types

Goal: To locate known cell types within the dataset.

目標(biāo)：查找數(shù)據(jù)集中已知細(xì)胞類型。

Identifying cell types from known markers is straightforward and fast in Loupe Cell Browser. We'll do this two ways, first through looking at quantitative accessibility, then by importing feature lists.

使用Loupe Cell Browser根據(jù)已知markers識(shí)別細(xì)胞類型非常直接形入，快速全跨。我們將通過(guò)兩種方式實(shí)現(xiàn)，首先是關(guān)注量化的可接近性亿遂，接著是通過(guò)導(dǎo)入特征列表浓若。

特征搜索 - Feature Search

Let's get a feel for the t-SNE plot for this dataset. We're looking at a PBMC sample, so we would hope to see relatively clear clustering of common cell types. As every t-SNE plot is different, we need to use feature markers to orient ourselves. Let's start by using promoter markers to find cellular types.

我們首先來(lái)感受一下這個(gè)數(shù)據(jù)集的t-SNE圖渺杉。我們正在查看一個(gè)PBMC樣本，因此我們希望能夠相對(duì)清晰的看到常見(jiàn)細(xì)胞類型分群挪钓。由于每一個(gè)t-SNE圖都是不一樣的是越，我們需要利用特征標(biāo)記(feature markers)來(lái)為我們定向。讓我們先以啟動(dòng)子標(biāo)記來(lái)尋找細(xì)胞類型作為開始吧碌上。

利用啟動(dòng)子總和確定細(xì)胞類型 - Using Promoter Sums to Determine Cell Type

The Cell Ranger ATAC pipeline labels individual peaks as promoters for a particular gene if the peak falls 1000 bases upstream from a gene's transcription start site, or 100 bases downstream from a gene's transcription start site. A promoter sum for a given gene is the number of cut sites per cell that fall within all the peaks labeled as promoters for that gene. Even though most cell type specific chromatin accessibility happens distal from promoters (a well known phenomenon that we observe in our data), we have found that promoter sums are sufficient to identify cell subtypes, as genes known to be upregulated in those subtypes will likely be more accessible at the promoter. Let's use some T-cell and B-cell gene markers to test this.

如果peak落在一個(gè)基因的轉(zhuǎn)錄起始位點(diǎn)上下游1000bp范圍內(nèi)倚评，Cell Ranger ATAC流程就將單個(gè)(?)peaks標(biāo)記為特定基因的啟動(dòng)子。對(duì)于一個(gè)給定基因的啟動(dòng)子總和就是落在該基因所有被標(biāo)記為啟動(dòng)子的peaks內(nèi)的每個(gè)細(xì)胞切割位點(diǎn)的數(shù)量馏予。盡管大多數(shù)細(xì)胞類型特異性的染色質(zhì)可接近性發(fā)生在啟動(dòng)子遠(yuǎn)端（我們?cè)谖覀兊臄?shù)據(jù)中觀測(cè)到的一個(gè)常見(jiàn)現(xiàn)象）天梧，我們發(fā)現(xiàn)由于這些亞型中已知會(huì)上調(diào)的基因在啟動(dòng)子區(qū)域更傾向于可接近，啟動(dòng)子總和足以用來(lái)識(shí)別細(xì)胞亞型霞丧。讓我們用一些T細(xì)胞和B細(xì)胞的基因markers來(lái)檢驗(yàn)一下呢岗。

First, select Accessibility mode from the Mode Selector. You will see an Active Feature List. This is like a scratch pad for exploring markers and motifs in your dataset. You can type in the name of a gene, transcription factor motif, or even a peak genomic region into the search box. Let's look for B cells first. Type MS4A1 (CD20) into the search box to bring up the "MS4A1 Sum" feature. Press Tab or Enter to add the promoter sum to the active feature list, and calculate cut site counts for that promoter across the dataset.

首先，在模式選擇器中選擇可接近性模式蛹尝。你會(huì)看得到一個(gè)活動(dòng)特征列表后豫。在你探索你數(shù)據(jù)集中的markers和motifs時(shí)，這就像一個(gè)便簽本一樣箩言。你可以在搜索框中輸入一個(gè)基因的名稱硬贯，轉(zhuǎn)錄因子基序，或者是一個(gè)peak的基因組區(qū)域陨收。讓我們先來(lái)看一下B細(xì)胞饭豹。在搜索框中輸入MS4A1(CD20)，檢索得到“MS4A1 Sum”特征务漩。按下Tab鍵或回車鍵來(lái)將啟動(dòng)子總和加入活動(dòng)特征列表拄衰，并計(jì)算整個(gè)數(shù)據(jù)集中該啟動(dòng)子的切割位點(diǎn)數(shù)。

atac_bcell.png

Right away, you should see that cells where the MS4A1 transcription start site was accessible are neatly packed into one of the t-SNE clusters. To confirm, you can try some other B-cell markers: CD19 and IGKC. It is clear that distinct region represents B cells.

緊接著饵骨，你應(yīng)該看到那些MS4A1轉(zhuǎn)錄起始位點(diǎn)可接近的細(xì)胞被整潔的劃歸到了t-SNE分群之一翘悉。為了確認(rèn)，你可以嘗試一些其他的B細(xì)胞markers：CD19和IGKC居触。很明顯那個(gè)獨(dú)特的區(qū)域代表B細(xì)胞妖混。

As you add multiple features to a list, the coloring will represent a combination of cut site counts per cell among all features in the list. With MS4A1, CD19 and IGKC in the list and 'Feature Max' as the selected attribute, the coloring of the plot reflects the maximum cut site count per cell among the selected features. Clicking on a single peak or promoter sum within a list shows the number of cut site counts per cell for that feature. Finally, hovering over a promoter sum or peak will reveal a graph and a trash icon; clicking the graph icon will highlight that feature in the peak viewer (see below), and clicking the trash icon will remove that feature from the current list.

當(dāng)你向一個(gè)列表中加入多個(gè)特征時(shí)，著色將表示列表中所有特征的每細(xì)胞切割位點(diǎn)計(jì)數(shù)的組合轮洋。當(dāng)使用MS4A1制市，CD19和IGKC特征，并選擇“Feature Max”屬性弊予。圖的著色反映了所選特征中每細(xì)胞最大切割點(diǎn)計(jì)數(shù)祥楣。單擊一個(gè)列表內(nèi)的一個(gè)peak或啟動(dòng)子總和將展示該特征的每細(xì)胞切割位點(diǎn)計(jì)數(shù)的數(shù)量。最后，鼠標(biāo)懸停在一個(gè)啟動(dòng)子總和或peak將顯示一個(gè)圖標(biāo)和一個(gè)垃圾箱圖標(biāo)误褪；單擊圖表圖標(biāo)將在peak viewer中高亮該特征（見(jiàn)下文）责鳍，單擊垃圾箱圖標(biāo)將在當(dāng)前列表中刪除該特征。

bcell_peakviewer.png

利用轉(zhuǎn)錄因子motifs確定細(xì)胞類型 - Using Transcription Factor Motifs to Determine Cell Type

Transcription factor motif patterns may also yield insights about cell type. Motifs are different from peak and promoter sums. For each transcription factor motif, Cell Ranger computes a z-score for each barcode, which represents the relative accessibility of all peaks containing that motif within that cell, compared to the entire dataset. When selected, motif z-scores are displayed in the t-SNE plot, rather than absolute cut site counts. In addition, motifs may only be selected one at a time.

轉(zhuǎn)錄因子motif模式也可以獲得一些細(xì)胞類型方面的信息兽间。Motifs不同于peak和啟動(dòng)子總和历葛。對(duì)于每一個(gè)轉(zhuǎn)錄因子motif，Cell Ranger對(duì)每一個(gè)barcode計(jì)算一個(gè)z-score渡八，這些z-score代表了與整個(gè)數(shù)據(jù)集相比啃洋，一個(gè)細(xì)胞中所有包括該motif的peaks的相對(duì)可接近性。

Let's reset the Active Feature List by clicking the trash icon for each feature currently in the list. Next, select SPI1 from the feature search box. The SPI1 (PU.1) transcription factor has been shown to play a key role in monocyte function [1]. Selecting SPI1 shows the cells with higher z-scores in red, indicating higher relative accessibility among all peaks that have the SPI1 motif. SPI1 also has a role in B-cell regulation; since we already identified the B-cell cluster through gene markers, it should follow that the large cluster at upper left are monocytes.

讓我們通過(guò)點(diǎn)擊垃圾箱圖標(biāo)刪除當(dāng)前列表中的所有特征來(lái)重置活動(dòng)特征列表屎鳍。接下來(lái)宏娄，在特征搜索框中選擇SPI1。研究表明逮壁，SPI1(PU.1)轉(zhuǎn)錄因子在單核細(xì)胞功能中扮演了關(guān)鍵角色[1]孵坚。選擇SPI1將高z-score的細(xì)胞顯示為紅色，表明在所有具有SPI1 motif的peak之間具有較高的相對(duì)可達(dá)性窥淆。SPI1在B細(xì)胞調(diào)節(jié)中也具有一定作用卖宠；由于我們已經(jīng)通過(guò)基因標(biāo)記識(shí)別了B細(xì)胞分群，因此忧饭，左上方的大群應(yīng)該是單核細(xì)胞扛伍。

spi1.png

導(dǎo)入和導(dǎo)出特征列表 - Importing and Exporting Feature Lists

Another way to find cell subtypes is to import a CSV file which contains markers for cell types of interest. You can download one such file here: ATACBloodCell.csv. After downloading, click on the three dots to the right of the current feature list, and select 'Import Lists' from the dropdown menu. Select the file you just downloaded. You can find more information about how to generate feature lists on the Sharing Results page.

另一種發(fā)現(xiàn)細(xì)胞亞型的方式是導(dǎo)入一個(gè)包含所有感興趣細(xì)胞類型marker的CSV文件。你可以從這里下載一個(gè)這樣的文件：ATACBloodCell.csv词裤。下載后刺洒，點(diǎn)擊當(dāng)前特征列表右上方的三個(gè)點(diǎn)按鈕，在彈出菜單中選擇“導(dǎo)入列表”吼砂。選擇你剛剛下載好的文件逆航。你可以在分享結(jié)果部分找到更多關(guān)于如何生成特征列表的信息。

import-list.png

After import, you should now be able to select from one of five sets of markers by selecting from the feature list selector (click the toggle next to the current feature list).

導(dǎo)入之后渔肩，現(xiàn)在你應(yīng)該能夠通過(guò)從特征列表選擇器（單擊當(dāng)前特征列表旁邊的切換按鈕）中選擇五組標(biāo)記之一了因俐。

You can use the feature list menu to create and export your own sets of markers, and rename and delete lists from your workspace. Choosing Export Lists from the feature menu will write all the currently active lists in your dataset to a CSV file, which you can import into other datasets. NOTE: If you wish to save the features in the Active Feature List, be sure to rename the list to something else prior to export.

你可以使用特征列表菜單來(lái)創(chuàng)建和導(dǎo)出你自己的marker集，在你的工作空間中重命名及刪除列表周偎。從功能菜單中選擇導(dǎo)出列表將把數(shù)據(jù)集中所有當(dāng)前活動(dòng)的列表寫到CSV文件中抹剩，你可以將該文件導(dǎo)入到其他數(shù)據(jù)集中。注意：如果你希望保存活動(dòng)特征列表中的特征蓉坎，請(qǐng)務(wù)必將列表在到處之前重命名為任意其他名稱吧兔。

保存細(xì)胞類型分群 - Saving Cell Type Clusters

Now that we've gotten our bearings in the dataset, we can save cell types for later. We can create new clusters corresponding to our cell types either by manual selection, or by quantitative filtering.

現(xiàn)在我們已經(jīng)在數(shù)據(jù)集中找到了（感興趣細(xì)胞類型）大概的位置，我們可以保存細(xì)胞類型以便后續(xù)使用袍嬉。我們可以通過(guò)人工選擇或根據(jù)定量過(guò)濾來(lái)創(chuàng)建新的細(xì)胞類型分群。

Let's first create a B-cell cluster. Select the rectangular lasso tool from the toolbox, and drag a box around the cluster we found through B-cell marker (MS4A1, CD19) accessibility:

讓我們先來(lái)創(chuàng)建一個(gè)B細(xì)胞分群。在工具箱中選擇矩形套索工具伺通，拖動(dòng)選框選中我們通過(guò)B細(xì)胞marker(MS4A1, CD19)可接近性發(fā)現(xiàn)的分群：

select.png

When you finish dragging, a dialog box will appear, prompting you to type the name of a new or existing category, and a cluster name. Create a new "Cell Types" category, and call the cells "B Cells". Press the Save button, and a new Cell Types category will appear.

當(dāng)你完成選中后箍土，會(huì)彈出一個(gè)對(duì)話框，提示你輸入一個(gè)新的或已存在的分類名稱罐监，以及一個(gè)分群名稱吴藻。創(chuàng)建一個(gè)新的“細(xì)胞類型”分類，將這些細(xì)胞命名為“B細(xì)胞”弓柱。按下保存按鈕沟堡，一個(gè)新的細(xì)胞類型分類就創(chuàng)建成功了。

Next, use the freehand lasso tool from the toolbox to draw an area around the monocytes, at the upper left of the t-SNE projection. Reuse the Cell Types category by typing it in the Category box, and create a new "Monocytes" cluster.

接下來(lái)矢空，在工具箱中選擇自由套索工具在t-SNE投射圖左上方選中單核細(xì)胞區(qū)域航罗。在分類框中選擇輸入細(xì)胞類型，創(chuàng)建一個(gè)新的“單核細(xì)胞”分群屁药。

You may also create clusters quantitatively. Switch back to Accessibility mode, and select the All T Cells list that was just imported from ATACBloodCell.csv, or choose "CD3D Sum" from the feature search box. Click on the CD3D Sum feature in the list, and then find the input box under "Select by Count - CD3D Sum". Enter zero, and then press the filter button next to the input field. This will highlight every cell for which there was a fragment within a CD3D promoter peak, and bring up the cluster assignment box. Select "Cell Types" as the category, and add these cells to the "T Cells" cluster.

你也可以定量的創(chuàng)建分群粥血。切換回可接近性模式，選擇剛剛從 ATACBloodCell.csv 中導(dǎo)入的全部T細(xì)胞酿箭，或在特征搜索框中選擇“CD3D Sum”特征复亏。單擊列表中的CD3D Sum特征，接著找到“Select by Count - CD3D Sum”下方的輸入框缭嫡。輸入“0”缔御，接著點(diǎn)擊輸入框旁邊的過(guò)濾按鈕。這將會(huì)高亮CD3D啟動(dòng)子peak內(nèi)有fragment的每一個(gè)細(xì)胞妇蛀，并彈出指定分群對(duì)話框耕突。選擇“細(xì)胞類型”作為分類，將這些細(xì)胞加入到“T細(xì)胞”分群中讥耗。

We'll stop here, but you can use additional genes and motifs of interest to further divide T cells into cytotoxic T cells, helper T cells, and memory T cells.

到這里我們將告一段落有勾，當(dāng)然你可以利用其他感興趣的基因或者M(jìn)otifs來(lái)進(jìn)一步的將T細(xì)胞劃分為殺傷性T細(xì)胞，輔助性T細(xì)胞古程，以及記憶T細(xì)胞蔼卡。

tcell-select.png

Before proceeding, please save these cluster assignments by clicking on the Save icon in the toolbar. The ATAC Tutorial file bundled with Loupe Cell Browser is read-only, so you will be prompted to save a copy somewhere on your file system.

在進(jìn)一步開始前，請(qǐng)點(diǎn)擊工具箱中的保存按鈕來(lái)保存這些指定好的分群挣磨。Loupe Cell Browser附帶的ATAC教程文件的權(quán)限為只讀雇逞，因此你將被提示將文件另存為其他副本來(lái)保存到系統(tǒng)上。

With feature markers and clusters saved, let's delve further into this ATAC data through the Peak Viewer.

保存好特征markers及分群后茁裙，讓我們使用Peak Viewer進(jìn)一步深入挖掘ATAC數(shù)據(jù)塘砸。

[1] https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0033474

結(jié)束語(yǔ)

感謝您閱讀到此處，本文由Cerasus_sp翻譯晤锥，浩渺予懷校對(duì)掉蔬。如果您喜歡我們的文章廊宪，請(qǐng)贊賞鼓勵(lì)！