Seurat - Guided Clustering Tutorial, July 16, 2020

https://satijalab.org/seurat/v3.2/pbmc3k_tutorial.html

下载数据：

https://cf.10xgenomics.com/samples/cell/pbmc3k/pbmc3k_filtered_gene_bc_matrices.tar.gz

需要这三个文件：

barcodes.tsv、genes.tsv、matrix.mtx

本次的数据分析流程，选择了网上10X Genomics上 PBMC 外周血单核细胞（Peripheral Blood Mononuclear Cells ）的公开数据。这里使用了 Illumina NextSeq 500仪器测序了 2700 个细胞。raw data可以在这里找到（https://cf.10xgenomics.com/samples/cell/pbmc3k/pbmc3k_filtered_gene_bc_matrices.tar.gz）。

   【Read10X】函数读取10X  [cellranger](https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/what-is-cell-ranger) pipeline的输出文件，返回一个 UMI count matrix（unique molecular identified）。在这个matrix里面的数值代表了每个特征(feature)（例如基因，行）的分子数目，这些特征可在每个细胞（列）里面被检测到。

     接着我们使用count matrix去创造一个【Seurat】对象。这个对象被用作单细胞数据集的，装载数据（类似count matrix）和分析（类似PCA，或聚类结果）的容器。count matrix 被储存在【pbmc[["RNA"]]@counts】。

rm(list=ls())

options(stringsAsFactors = F)

library(dplyr)

library(Seurat)

library(patchwork)

Load the PBMC dataset

pbmc.data <- Read10X(data.dir = "filtered_gene_bc_matrices/hg19/")

image.png

Environment

History

Connections

List

mportDatasct

GlobalEnvironment

Data

LargedgcMatrix(88392600elements,.

pbmc.data

[1:22868841701661783263634104

0412492494495

.ai:int

[1:2701]

78121333264422447465528631171017634

int

..ap

[1:2]327382700

.@DiM:int

.@Dimnames:Listof2

.S:chr[1:32738]"MIR1302-10""AM138A""R4FS""P11-34P13.7"

AAACATT...

..:chr[1:2700]"AAACATACAACCAC-1""AAACATTGAGCTAC-1"

.x:huM[1:2286884]11211114111.

afactors:listO

Initialize the Seurat object with the raw (non-normalized data).

创建Seurat对象

pbmc <- CreateSeuratObject(counts = pbmc.data, project = "pbmc3k", min.cells = 3, min.features = 200)

Warning: Feature names cannot have underscores ('_'), replacing with dashes ('-')

pbmc

An object of class Seurat

13714 features across 2700 samples within 1 assay

Active assay: RNA (13714 features, 0 variable features)

Standard pre-processing workflow

质量控制指标（QC metrics）和筛选细胞作进一步分析：

Seurat允许根据任何用户定义的标准过滤单元格（cells）。

通常使用的QC指标：

每个细胞内被检测到特有的基因（****unique genes****）的数目，unique feature会因为数据质量而调整。

▲低质量的细胞或空的液滴一般含有较少的基因；

▲细胞双重态或多重态可能呈现异常高的基因count值

细胞内被监测到的分子的总数目（与unique genes高度相关）

匹配到线粒体基因组的read的百分比

▲低质量/将要死去的细胞经常呈现过度的线粒体污染情况；

▲使用

function计算线粒体QC metrics, 此function可以计算源于feature集的count值的百分比
▲使用所有以【MT-】为起始的基因集合，作为线粒体基因集
The [[ operator can add columns to object metadata. This is a great place to stash QC stats

pbmc[["percent.mt"]] <- PercentageFeatureSet(pbmc, pattern = "^MT-")

QC metrics 在 Seurat ：head([email protected])) QC指标存储

▲过滤掉拥有 feature count>2500, or < 200 的细胞；

▲过滤掉含有> 5% 的线粒体 count值的细胞

Visualize QC metrics as a violin plot

nCount RNA: 这些基因数目一共测到的count数目,也就是以前版本的UMI数目

nFeature RNA: 每个细胞所检测到的基因数目，也就是以前版本的nGene

percent.mt: 每个细胞所检测到的线粒体基因

VlnPlot(pbmc, features = c("nFeature_RNA", "nCount_RNA", "percent.mt"), ncol = 3)

image.png

nFeature_RNA

nCountRNA

percent.mt

15000

3000

10000

2000

5000

1000

ouo

ldentity

Identity

FeatureScatter is typically used to visualize feature-feature relationships, but can be used

for anything calculated by the object, i.e. columns in object metadata, PC scores etc.

plot1 <- FeatureScatter(pbmc, feature1 = "nCount_RNA", feature2 = "percent.mt")

plot2 <- FeatureScatter(pbmc, feature1 = "nCount_RNA", feature2 = "nFeature_RNA")

plot1 + plot2

image.png

-0.13

0.95

3000

NEeJnee-u

wtuasjad

2000

ldentity

pbmc3k

1000

15000

5000

10000

5000

15000

10000

nCountRNA

pbmc <- subset(pbmc, subset = nFeature_RNA > 200 & nFeature_RNA < 2500 & percent.mt < 5)

pbmc

An object of class Seurat

13714 features across 2638 samples within 1 assay

Active assay: RNA (13714 features, 0 variable features)

Normalizing the data

在去除了数据集中不需要的细胞后，下一步就是标准化数据。默认设置下，使用全局-规模（global-scaling）的标准化方法 "LogNormalize"，这方法按总体表达量，对每一个细胞的特征表达量进行标准化，乘以一个比例因子（scale factor）（默认为10,000），然后log转换此结果。

标准化后的值储存在【pbmc[["RNA"]]@data】。

pbmc <- NormalizeData(pbmc, normalization.method = "LogNormalize", scale.factor = 10000)

Performing log-normalization

0% 10 20 30 40 50 60 70 80 90 100%

[----|----|----|----|----|----|----|----|----|----|

▶ 为了更为清晰，在之前的代码（和之后的代码）中，他们在函数调用（function call）中给确定的参数设置提供了默认值。然而，这些设置不是必须的，而且同样的结果可由以下代码实现：

pbmc <- NormalizeData(pbmc)

Performing log-normalization

0% 10 20 30 40 50 60 70 80 90 100%

[----|----|----|----|----|----|----|----|----|----|

**|

Identification of highly variable features (feature selection)

下一步，计算数据集里面基因子集，该子集呈现出高度的细胞间变异（在有些细胞内高表达，在其他细胞内低表达）。还发现了专注在下游分析内的基因可帮助突出单细胞数据内的生物信号。

Seurat3 ，根据旧有版本升级了一下，通过使用【FindVariableFeatures 】函数，直接对单细胞数据中固有内的均值方差（mean-variance）的关系建模。在默认设置下，每个数据集返回2,000个feature。这些数据会被用于下游数据分析，如PCA。

pbmc <- FindVariableFeatures(pbmc, selection.method = "vst", nfeatures = 2000)

Calculating gene variances

0% 10 20 30 40 50 60 70 80 90 100%

[----|----|----|----|----|----|----|----|----|----|

**|

Calculating feature variances of standardized and clipped values

0% 10 20 30 40 50 60 70 80 90 100%

[----|----|----|----|----|----|----|----|----|----|

**|

Identify the 10 most highly variable genes

top10 <- head(VariableFeatures(pbmc), 10)

top10

[1] "PPBP" "LYZ" "S100A9" "IGLL5" "GNLY" "FTL" "PF4" "FTH1"

[9] "GNG11" "S100A8"

plot variable features with and without labels

plot1 <- VariableFeaturePlot(pbmc)

plot2 <- LabelPoints(plot = plot1, points = top10, repel = TRUE)

When using repel, set xnudge and ynudge to 0 for optimal results

plot1 + plot2

Warning messages:

1: Transformation introduced infinite values in continuous x-axis

2: Transformation introduced infinite values in continuous x-axis

image.png

PPBP

S100A9

LYZ

IGLL5

aoupieApezipjepueis

oueueApazipjepueis

GNLY

FTH1

PF4

GNG11S100A8

Non-yariablecount11714

Non-variablecount11714

Variablecount:2000

Variablecount2000

1e-021e+001e+02

1e-021e+001e+02

AverageExpression

AverageExpression

Scaling the data

使用线性转换('scaling')，它是一个降维技巧（如PCA）的标准预处理步骤，【ScaleData】函数：

▲转换每个基因的表达量，使得细胞之间的平均表达量为0；

▲按比例缩小每个基因的表达量，使得细胞之间的方差（variance）为1；

   ☼此步骤给予了下游分析的同等权重（equal weight），使得高表达基因不会占主要地位；

▲改结果会被储存在【pbmc[["RNA"]]@scale.data】

all.genes <- rownames(pbmc)

pbmc <- ScaleData(pbmc, features = all.genes)

Centering and scaling data matrix

|=======================================================================| 100%

▶如果这一步骤太久了，可以这么加快？

按比例缩小数据在Seurat的工作路线里面是必须的一步，但仅在基因这方面上会被用作输入数据。因此，【ScaleData】内的默认设置仅仅是为了对之前定义好的变量特征进行按比例缩小（默认为2000个基因）。为了进行上一步，可删掉类似之前函数调用里的【features】参数。

pbmc <- ScaleData(pbmc)

Centering and scaling data matrix

|==========================================================| 100%

pbmc

An object of class Seurat

13714 features across 2638 samples within 1 assay

Active assay: RNA (13714 features, 2000 variable features)

2 dimensional reductions calculated: pca, umap

▶应该怎么去除一些不需要的变量来源，就像在Seurat v2那样？

在【Seurat v2】里面，我们也会使用【ScaleData】函数来去除单细胞数据集里面一些不需要的变量来源。例如，我们可以“退回（regress out）”与例如细胞周期阶段相关的异质性，或与线粒体污染相关的异质性。这些特征会在【Seurat v3】中的【ScaleData】中得到认证。

pbmc <- ScaleData(pbmc, vars.to.regress = "percent.mt")

Regressing out percent.mt

|| 100%
Centering and scaling data matrix
|| 100%

pbmc

An object of class Seurat

13714 features across 2638 samples within 1 assay

Active assay: RNA (13714 features, 2000 variable features)

2 dimensional reductions calculated: pca, umap

Perform linear dimensional reduction

对scaled过的数据进行PCA分析。默认设置上，只有之前的确定变量的feature被用于输入数据，但可被【features】参数定义，如果你希望选择一个不同的子集。

pbmc <- RunPCA(pbmc, features = VariableFeatures(object = pbmc))

PC_ 1

Positive: CST3, TYROBP, LST1, AIF1, FTL, FTH1, LYZ, FCN1, S100A9, TYMP

FCER1G, CFD, LGALS1, S100A8, CTSS, LGALS2, SERPINA1, IFITM3, SPI1, CFP

PSAP, IFI30, SAT1, COTL1, S100A11, NPC2, GRN, LGALS3, GSTP1, PYCARD

Negative: MALAT1, LTB, IL32, IL7R, CD2, B2M, ACAP1, CD27, STK17A, CTSW

CD247, GIMAP5, AQP3, CCL5, SELL, TRAF3IP3, GZMA, MAL, CST7, ITM2A

MYC, GIMAP7, HOPX, BEX2, LDLRAP1, GZMK, ETS1, ZAP70, TNFAIP8, RIC3

PC_ 2

Positive: CD79A, MS4A1, TCL1A, HLA-DQA1, HLA-DQB1, HLA-DRA, LINC00926, CD79B, HLA-DRB1, CD74

HLA-DMA, HLA-DPB1, HLA-DQA2, CD37, HLA-DRB5, HLA-DMB, HLA-DPA1, FCRLA, HVCN1, LTB

BLNK, P2RX5, IGLL5, IRF8, SWAP70, ARHGAP24, FCGR2B, SMIM14, PPP1R14A, C16orf74

Negative: NKG7, PRF1, CST7, GZMB, GZMA, FGFBP2, CTSW, GNLY, B2M, SPON2

CCL4, GZMH, FCGR3A, CCL5, CD247, XCL2, CLIC3, AKR1C3, SRGN, HOPX

TTC38, APMAP, CTSC, S100A4, IGFBP7, ANXA1, ID2, IL32, XCL1, RHOC

PC_ 3

Positive: HLA-DQA1, CD79A, CD79B, HLA-DQB1, HLA-DPB1, HLA-DPA1, CD74, MS4A1, HLA-DRB1, HLA-DRA

HLA-DRB5, HLA-DQA2, TCL1A, LINC00926, HLA-DMB, HLA-DMA, CD37, HVCN1, FCRLA, IRF8

PLAC8, BLNK, MALAT1, SMIM14, PLD4, LAT2, IGLL5, P2RX5, SWAP70, FCGR2B

Negative: PPBP, PF4, SDPR, SPARC, GNG11, NRGN, GP9, RGS18, TUBB1, CLU

HIST1H2AC, AP001189.4, ITGA2B, CD9, TMEM40, PTCRA, CA2, ACRBP, MMD, TREML1

NGFRAP1, F13A1, SEPT5, RUFY1, TSC22D1, MPP1, CMTM5, RP11-367G6.3, MYL9, GP1BA

PC_ 4

Positive: HLA-DQA1, CD79B, CD79A, MS4A1, HLA-DQB1, CD74, HLA-DPB1, HIST1H2AC, PF4, TCL1A

SDPR, HLA-DPA1, HLA-DRB1, HLA-DQA2, HLA-DRA, PPBP, LINC00926, GNG11, HLA-DRB5, SPARC

GP9, AP001189.4, CA2, PTCRA, CD9, NRGN, RGS18, GZMB, CLU, TUBB1

Negative: VIM, IL7R, S100A6, IL32, S100A8, S100A4, GIMAP7, S100A10, S100A9, MAL

AQP3, CD2, CD14, FYB, LGALS2, GIMAP4, ANXA1, CD27, FCN1, RBP7

LYZ, S100A11, GIMAP5, MS4A6A, S100A12, FOLR3, TRABD2A, AIF1, IL8, IFI6

PC_ 5

Positive: GZMB, NKG7, S100A8, FGFBP2, GNLY, CCL4, CST7, PRF1, GZMA, SPON2

GZMH, S100A9, LGALS2, CCL3, CTSW, XCL2, CD14, CLIC3, S100A12, CCL5

RBP7, MS4A6A, GSTP1, FOLR3, IGFBP7, TYROBP, TTC38, AKR1C3, XCL1, HOPX

Negative: LTB, IL7R, CKB, VIM, MS4A7, AQP3, CYTIP, RP11-290F20.3, SIGLEC10, HMOX1

PTGES3, LILRB2, MAL, CD27, HN1, CD2, GDI2, ANXA5, CORO1B, TUBA1B

FAM110A, ATP1A1, TRADD, PPA1, CCDC109B, ABRACL, CTD-2006K23.1, WARS, VMO1, FYB

pbmc

An object of class Seurat

13714 features across 2638 samples within 1 assay

Active assay: RNA (13714 features, 2000 variable features)

1 dimensional reduction calculated: pca

Seurat提供了几个有用的，决定 PCA 的可视化细胞和feature的方法，包括：【VizDimReduction】、【DimPlot】、【DimHeatmap】

Examine and visualize PCA results a few different ways

print(pbmc[["pca"]], dims = 1:5, nfeatures = 5)

PC_ 1

Positive: CST3, TYROBP, LST1, AIF1, FTL

Negative: MALAT1, LTB, IL32, IL7R, CD2

PC_ 2

Positive: CD79A, MS4A1, TCL1A, HLA-DQA1, HLA-DQB1

Negative: NKG7, PRF1, CST7, GZMB, GZMA

PC_ 3

Positive: HLA-DQA1, CD79A, CD79B, HLA-DQB1, HLA-DPB1

Negative: PPBP, PF4, SDPR, SPARC, GNG11

PC_ 4

Positive: HLA-DQA1, CD79B, CD79A, MS4A1, HLA-DQB1

Negative: VIM, IL7R, S100A6, IL32, S100A8

PC_ 5

Positive: GZMB, NKG7, S100A8, FGFBP2, GNLY

Negative: LTB, IL7R, CKB, VIM, MS4A7

VizDimLoadings(pbmc, dims = 1:2, reduction = "pca")

image.png

CD79A

TYi

48

LST

HLA-DOA1

HLA-DQB1

HLA-DRA

LINC00926

FCN1

CD79B

APMAP

S100A9

ITC38

TYMP

FCER1G

HOPX

CFD

SRGN

LGALS1

AKR1C3

CLIC3

S100A8

CTSS

品

CD247-

CCL5

IFITM3

FCGR3A

p

GZMH

CCL4

PSAP

SPON2

IFI30

RE

C8L1

CTSW

S100A11

FGFBP2

NPC2

GZMA

GRN

GZMB

L8SP

CST

MALAT11

0.05

.0.1

0.05

0.0

0.10

0.00

0.10

PC_1

PC2

DimPlot(pbmc, reduction = "pca")

image.png

10

口

pbmc3k

10

PC_1

可使用【DimHeatmap】对数据集内的异质性的主要数据作简单探索，在决定哪个PC可包含在内被用作下有分析时，此函数显得特别有用。细胞和feature根据PCA的打分值来设置。设置【cells】为一数值绘制出范围两端的“极端”细胞，这可显著地加速大数据集图表的绘制。尽管这是一监督算法分析，发现这仍是用于探索相关feature特征的有价值的工具。

DimHeatmap(pbmc, dims = 1, cells = 500, balanced = TRUE)

image.png

MALAT1

LTB

1L32

IL7R

CD2

B2M

ACAP1

CD27

STK17A

CTSW

CD247

GILMAP5

SELL

CTSS

S100A8

LGALS1

CFD

FCER1G

TYMP

S100A9

FCN1

TYROBP

CST3

DimHeatmap(pbmc, dims = 1:15, cells = 500, balanced = TRUE)

image.png

5100A

PC5

PC6

CDM

PC8

PC9

cLEC10

ICB9

WMCA2

PPPRY

UAO1

PC11

PC14

民间

Determine the 'dimensionality' of the dataset

为了克服scRNA-seq数据里面单一特征的广泛技术性噪音，Seurat是基于它们的 PCA 分数聚集细胞的，每一个PC 本质上而言代表了“metafeature”，也就是那些包含相关特征数据集的信息。因此，前几个主要的成分代表了数据集强而有力的压缩。问题是，我们应该囊括多少个成分在内呢？10个? 20个？100个？

在 Macosko的文章内，使用了受JackStraw 程序启发的重新取样测试（ resampling test ）。随机重新排列了数据集（默认设置为1%），然后再次跑了 PCA，重建特征打分值（feature scores）下的零分布（null distribution），再重复此步骤。鉴定了“显著的”PC作为那些拥有低P-value特征的强有力的富集。

NOTE: This process can take a long time for big datasets, comment out for expediency. More

approximate techniques such as those implemented in ElbowPlot() can be used to reduce

computation time

pbmc <- JackStraw(pbmc, num.replicate = 100)

|| 100% elapsed=02m 48s

|| 100% elapsed=00s

pbmc <- ScoreJackStraw(pbmc, dims = 1:20)

pbmc

An object of class Seurat

13714 features across 2638 samples within 1 assay

Active assay: RNA (13714 features, 2000 variable features)

1 dimensional reduction calculated: pca

【JackStrawPlot】函数为每一个PC的P-value的分布与一均匀分布（ uniform distribution）的对比提供了可视化的工具（虚线）。“显著的”PC会显示带有较低的P-value的特征的强富集（实线上的虚线）。在此情况下，在第一个10~12 PC之后，有一个显著的断崖式下降。

JackStrawPlot(pbmc, dims = 1:15)

Warning message:

Removed 23496 rows containing missing values (geom_point).

image.png

0.3

pC:p-yalue

PC1:9.12e-161

PC2:1.47e-114

[oooDunjleonajoayl

PC3:1e-32

PC4:3.04e-58

0.2

PC5:5.05e-56

PC6:1.57e-55

PC7:4.05e-24

PC8:1.48e-11

PC9:5.22e-12

PC10:6.33e-08

0.1

PC11:6.33e-08

PC12:0.000101

PC13:000253

PC14:0.133

PC15:1

0.0

0.075

0.000

0.100

0.025

0.050

Empirical

另一种启发式方法（heuristic method）可以生成一个“Elbow plot”：主成分的排列是基于差异的百分比，可其差异可通过【ElbowPlot】函数来解释。在此例子中，可观察到转折点位于PC9~10，这意味着真正信号的绝大多数会在第一个 10PC 处被捕捉。

ElbowPlot(pbmc)

image.png

uoneInenpjepueis

鉴定一个数据集的真正维度--对于使用者而言是一个挑战/未知之事。因此我们建议使用者去考虑着三个方法：

1、带有更多的监督的方法，探索PC 以用于决定异质性的相关来源，然后这可以用于类似与GSEA的共同使用上。

2、基于随机的零模型进行一统计学测试，但这对于大型数据而言是耗时的，而且不会返回清晰的 PC 截点（cutoff）。

3、启发式的方法，且是被广泛使用的，可被立马计算出来的。

在这个例子中，这三种方法都可产生相似的结果，但我们可能已经选择了任何位于PC 7-12的点作为合理的截点。

在这里，我们选择了10，但鼓励使用者考虑以下几点：

1、DC和NK的狂热粉可能认出了这些基因与PC12 、13高度相关，因此决定了稀有免疫亚群（例如MZB1是类浆细胞DC（plasmacytoid DCs）的一个 marker。然而这些群组太稀少，使得这样大小的数据集在不带有先前的认识下很难从背景噪音中被区分开。

2、我们鼓励使用者选择PC不同的数目重复下游分析（10、15、甚至50！），当你进行观察时，这结果一般不会显著差异过大。

3、我们建议使用者在选择这个参数时宁可多犯错。例如，仅带有5个PC 进行下游分析时也不会显著地、或负面影响结果。

Cluster the cells

Seurat v3 应用了以画图为基础的聚类方法，基于初始策略建立的(Macosko et al)。更为重要的是，距离度量使得聚类分析（基于之前鉴定出的PC）仍然维持一致。然而，我们的方法，即将细胞距离矩阵分区为小聚群，这一能力有所提升。我们的方法是受到近期的一手稿所启示，那个运用了对于scRNA seq数据 [SNN-Cliq, Xu and Su, Bioinformatics, 2015] 和CyTOF数据 [PhenoGraph, Levine et al., Cell, 2015]以画图为基础的聚类方法。简而言之，这些方法将细胞嵌入了图的结构内--如的k最近邻算法图 ( K-nearest neighbor graph****，KNN) ，其带有的edges被画在细胞之间，带有相似的特征表达模式，可被用于将一图划分为高度互通的“准小群体”或“社团”内。

在PhenoGraph里，在PCA的空间里面，我们第一次基于欧几里距离构建了KNN图，基于它们本地近邻之间共有的重叠部分，改善了任何两个细胞之间的边权（edge weights）。这一步通过【FindNeighbors】函数实现，且将先前定义好的数据集（起初10个PC）的维度作为输入。

为了聚类好细胞，我们下一步运用模块优化技术，类似默认的算法（Louvain algorithm）或SLM [SLM, Blondel et al., Journal of Statistical Mechanics]，以优化标准模块函数为目标地，迭代地将细胞分组。【FindClusters】函数使用了这一步骤，且包含了一个清晰度的参数，它可以设置下游聚类的间隔距离（granularity），使用上升的数值获得大量的集群。我们发现为越3000下拨的单细胞数据集设置这一参数约 0.4-1.2 ，典型地会返回一个好的结果。最优的分辨率一般会因为更大的数据集而升高。这集群可以通过【Idents】函数被找到。

pbmc <- FindNeighbors(pbmc, dims = 1:10)

Computing nearest neighbor graph

Computing SNN

pbmc <- FindClusters(pbmc, resolution = 0.5)

Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Ec

Number of nodes: 2638

Number of edges: 96033

Running Louvain algorithm...

0% 10 20 30 40 50 60 70 80 90 100%

[----|----|----|----|----|----|----|----|----|----|

**|

Maximum modularity in 10 random starts: 0.8720

Number of communities: 9

Elapsed time: 0 seconds

Look at cluster IDs of the first 5 cells

head(Idents(pbmc), 5)

AAACATACAACCAC-1 AAACATTGAGCTAC-1 AAACATTGATCAGC-1 AAACCGTGCTTCCG-1

1 3 1 2

AAACCGTGTATGCG-1

6

Levels: 0 1 2 3 4 5 6 7 8

Run non-linear dimensional reduction (UMAP/tSNE)

Seurat提供了多种非线性多降维的技巧，如 tSNE 和 UMAP，用于可视化和探索这些数据集。这些算法的目标是学习潜在的数据多样性为了将相似的细胞放置在低维度的空间。在上述以画图为基础的集群中的细胞应该会共存于这些降维图中。作为UMAP 和 tSNE 的输入数据，我们建议使用相同的 PC 作为聚类分群分析的输入数据。

If you haven't installed UMAP, you can do so via reticulate::py_install(packages ='umap-learn')

pbmc <- RunUMAP(pbmc, dims = 1:10)

Warning: The default method for RunUMAP has changed from calling Python UMAP via reticulate to the R-native UWOT using the cosine metric

To use Python UMAP via reticulate, set umap.method to 'umap-learn' and metric to 'correlation'

This message will be shown once per session

11:20:46 UMAP embedding parameters a = 0.9922 b = 1.112

11:20:46 Read 2638 rows and found 10 numeric columns

11:20:46 Using Annoy for neighbor search, n_neighbors = 30

11:20:46 Building Annoy index with metric = cosine, n_trees = 50

0% 10 20 30 40 50 60 70 80 90 100%

[----|----|----|----|----|----|----|----|----|----|

**|

11:20:46 Writing NN index file to temp file C:\Users\Dell\AppData\Local\Temp\RtmpU1dvCq\file306837978f5

11:20:46 Searching Annoy index using 1 thread, search_k = 3000

11:20:47 Annoy recall = 100%

11:20:47 Commencing smooth kNN distance calibration using 1 thread

11:20:47 Initializing from normalized Laplacian + noise

11:20:48 Commencing optimization for 500 epochs, with 105386 positive edges

0% 10 20 30 40 50 60 70 80 90 100%

[----|----|----|----|----|----|----|----|----|----|

**|

11:20:53 Optimization finished

note that you can set label = TRUE or use the LabelClusters function to help label

#individual clusters

DimPlot(pbmc, reduction = "umap")

image.png

10

NaiveCD4T

MemorvCD4T

CD14+Mono

三

CD8T

FCGR3A+Mono

NK

DC

Platelet

UMAP1

saveRDS(pbmc, file = "../pbmc_tutorial.rds")

Finding differentially expressed features (cluster biomarkers)

Seurat可帮助寻找那些通过差异表达决定分群的 markers。默认设置下，对比于其他的细胞，其鉴定了单一群落里面的正和负标记（positive and negative markers）（在指定的【ident.1】函数里）。【FindAllMarkers】函数自动化完成所有小集群的过程，但你可以测试小集群的分组，或其他细胞与细胞之间的，或针对所有细胞的分组。

【min.pct】参数或是在细胞的2个分组中的最小百分比处需要一个被检测到的特征，【thresh.test】参数需要一个两组之间存在一定差异表达量（通常而言）的特征。你可以设置这俩为0，但随着时间上的剧烈上升，由于这会检测到大量的特征，这些特征可能是高度无偏的。作为另一种去加快计算力的选择，可设置【max.cells.per.ident】。这会降低每个特征类别的取样，使其细胞数不超过设置的细胞数。然而一般而言，这会造成性能（power）的损失，这速度的提升可能会非常的显著，且大部分的高度差异显著特征将可能仍然上升至最大。

find all markers of cluster 1

cluster1.markers <- FindMarkers(pbmc, ident.1 = 1, min.pct = 0.25)

|++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed=01s

head(cluster1.markers, n = 5)

p_val avg_logFC pct.1 pct.2 p_val_adj

IL32 1.894810e-92 0.8373872 0.948 0.464 2.598542e-88

LTB 7.953303e-89 0.8921170 0.981 0.642 1.090716e-84

CD3D 1.655937e-70 0.6436286 0.919 0.431 2.270951e-66

IL7R 3.688893e-68 0.8147082 0.747 0.325 5.058947e-64

LDHB 2.292819e-67 0.6253110 0.950 0.613 3.144372e-63

find all markers distinguishing cluster 5 from clusters 0 and 3

cluster5.markers <- FindMarkers(pbmc, ident.1 = 5, ident.2 = c(0, 3), min.pct = 0.25)

|++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed=02s

head(cluster5.markers, n = 5)

p_val avg_logFC pct.1 pct.2 p_val_adj

FCGR3A 7.583625e-209 2.963144 0.975 0.037 1.040018e-204

IFITM3 2.500844e-199 2.698187 0.975 0.046 3.429657e-195

CFD 1.763722e-195 2.362381 0.938 0.037 2.418768e-191

CD68 4.612171e-192 2.087366 0.926 0.036 6.325132e-188

RP11-290F20.3 1.846215e-188 1.886288 0.840 0.016 2.531900e-184

find markers for every cluster compared to all remaining cells, report only the positive ones

pbmc.markers <- FindAllMarkers(pbmc, only.pos = TRUE, min.pct = 0.25, logfc.threshold = 0.25)

Calculating cluster 0

|| 100% elapsed=01s

Calculating cluster 1

|| 100% elapsed=00s

Calculating cluster 2

|| 100% elapsed=02s

Calculating cluster 3

|| 100% elapsed=01s

Calculating cluster 4

|| 100% elapsed=01s

Calculating cluster 5

|| 100% elapsed=02s

Calculating cluster 6

|| 100% elapsed=02s

Calculating cluster 7

|| 100% elapsed=02s

Calculating cluster 8

|++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed=01s

pbmc.markers %>% group_by(cluster) %>% top_n(n = 2, wt = avg_logFC)

Registered S3 method overwritten by 'cli':

method from

print.boxx spatstat

A tibble: 18 x 7

Groups: cluster [9]

p_val avg_logFC pct.1 pct.2 p_val_adj cluster gene

1 1.96e-107 0.730 0.901 0.594 2.69e-103 0 LDHB

2 1.61e- 82 0.922 0.436 0.11 2.20e- 78 0 CCR7

3 7.95e- 89 0.892 0.981 0.642 1.09e- 84 1 LTB

4 1.85e- 60 0.859 0.422 0.11 2.54e- 56 1 AQP3

5 0. 3.86 0.996 0.215 0. 2 S100A9

6 0. 3.80 0.975 0.121 0. 2 S100A8

7 0. 2.99 0.936 0.041 0. 3 CD79A

8 9.48e-271 2.49 0.622 0.022 1.30e-266 3 TCL1A

9 2.96e-189 2.12 0.985 0.24 4.06e-185 4 CCL5

10 2.57e-158 2.05 0.587 0.059 3.52e-154 4 GZMK

11 3.51e-184 2.30 0.975 0.134 4.82e-180 5 FCGR3A

12 2.03e-125 2.14 1 0.315 2.78e-121 5 LST1

13 7.95e-269 3.35 0.961 0.068 1.09e-264 6 GZMB

14 3.13e-191 3.69 0.961 0.131 4.30e-187 6 GNLY

15 1.48e-220 2.68 0.812 0.011 2.03e-216 7 FCER1A

16 1.67e- 21 1.99 1 0.513 2.28e- 17 7 HLA-DPB1

17 7.73e-200 5.02 1 0.01 1.06e-195 8 PF4

18 3.68e-110 5.94 1 0.024 5.05e-106 8 PPBP

Warning message:

... is not empty.

We detected these problematic arguments:

needs_dots

These dots only exist to allow future extensions and should be empty.

Did you misspecify an argument?

Seurat有多种差异表达的测试，它们可以在【test.use】参数内被设置。例如，ROC 测试会给任何独立的maker（范围由0-随机，到1-完美）返回一个“分类能力”。

cluster1.markers <- FindMarkers(pbmc, ident.1 = 0, logfc.threshold = 0.25, test.use = "roc", only.pos = TRUE)

|++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed=00s

我们涵盖了很多可视化marker表达量的工具。【VlnPlot】显示了集群里面表达量的概率分布，【FeaturePlot】（可用于可视化 tSNE 或 PCA图的特征表达量）是最经常使用的可视化工具。我们建议探索【 RidgePlot】、【CellScatter】和【DotPlot】作为另一种方法来探索数据集。

VlnPlot(pbmc, features = c("MS4A1", "CD79A"))

image.png

MS4A1

CD79A

[AUOISSEJDX3

AAETUOISseJdx3

Identity

you can plot raw counts as well

VlnPlot(pbmc, features = c("NKG7", "PF4"), slot = "counts", log = TRUE)

image.png

NKG7

PF4

AADUOISSAJDX3

en的uoissaJdx3

Identity

FeaturePlot(pbmc, features = c("MS4A1", "GNLY", "CD3E", "CD14", "FCER1A", "FCGR3A", "LYZ", "PPBP",
                           "CD8A"))

image.png

CD3E

GNLY

MS4A1

寸wweC

PC_1

PC1

FCER1A

CD14

FCGR3A

wwro

PC_1

LYZ

PPBP

CD8A

柯d

PC_1

【DoHeatmap】对给予的细胞和特征生成一热图。在这个情况下，我们对每个集群画出前20个marker（如果marker量少于20就全部的画上）

top10 <- pbmc.markers %>% group_by(cluster) %>% top_n(n = 10, wt = avg_logFC)

DoHeatmap(pbmc, features = top10$gene) + NoLegend()

image.png

州I国

HIST1H22

Assigning cell type identity to clusters

幸运的是，在这个数据集中，对于已知的细胞类型，我们可以使用权威的marker去轻易地吻合无偏倚的聚类

image.png

CellType

clusterID

Markers

IL7R,CCR7

NaiveCD4+T

1L7R.S100A4

MemoryCD4+

CD14+Mono

CD14,LYZ

MS4A1

CD8+T

CD8A

FCGR3A+Mono

FCGR3AMS4A7

GNLY.NKG7

FCER1ACST3

Platelet

PPBP

names(new.cluster.ids) <- levels(pbmc)

pbmc <- RenameIdents(pbmc, new.cluster.ids)

DimPlot(pbmc, reduction = "umap", label = TRUE, pt.size = 0.5) + NoLegend()

image.png

FCD14天MORO

2:0

三

Platelet

FCGR3A+MonN

CD8

-5

NateCD4.

Meno

CD45

-10

UMAP1

saveRDS(pbmc, file = "../pbmc3k_final.rds")

单细胞测序--Seurat包--官网经典流程

Seurat - Guided Clustering Tutorial, July 16, 2020

Load the PBMC dataset

Initialize the Seurat object with the raw (non-normalized data).

Standard pre-processing workflow

质量控制指标（QC metrics）和筛选细胞作进一步分析：

The [[ operator can add columns to object metadata. This is a great place to stash QC stats

QC metrics 在 Seurat ：head([email protected])) QC指标存储

Visualize QC metrics as a violin plot

FeatureScatter is typically used to visualize feature-feature relationships, but can be used

for anything calculated by the object, i.e. columns in object metadata, PC scores etc.

Normalizing the data

Identification of highly variable features (feature selection)

Identify the 10 most highly variable genes

plot variable features with and without labels

Scaling the data

▶如果这一步骤太久了，可以这么加快？

▶应该怎么去除一些不需要的变量来源，就像在Seurat v2那样？

Perform linear dimensional reduction

Examine and visualize PCA results a few different ways

Determine the 'dimensionality' of the dataset

NOTE: This process can take a long time for big datasets, comment out for expediency. More

approximate techniques such as those implemented in ElbowPlot() can be used to reduce

computation time

Cluster the cells

Look at cluster IDs of the first 5 cells

Run non-linear dimensional reduction (UMAP/tSNE)

If you haven't installed UMAP, you can do so via reticulate::py_install(packages ='umap-learn')

note that you can set `label = TRUE` or use the LabelClusters function to help label

#individual clusters

Finding differentially expressed features (cluster biomarkers)

find all markers of cluster 1

find all markers distinguishing cluster 5 from clusters 0 and 3

find markers for every cluster compared to all remaining cells, report only the positive ones

A tibble: 18 x 7

Groups: cluster [9]

you can plot raw counts as well

Assigning cell type identity to clusters

你可能感兴趣的:(单细胞测序--Seurat包--官网经典流程)

单细胞测序--Seurat包--官网经典流程

Seurat - Guided Clustering Tutorial, July 16, 2020

Load the PBMC dataset

Initialize the Seurat object with the raw (non-normalized data).

Standard pre-processing workflow

质量控制指标（QC metrics）和筛选细胞作进一步分析：

The [[ operator can add columns to object metadata. This is a great place to stash QC stats

QC metrics 在 Seurat ：head([email protected])) QC指标存储

Visualize QC metrics as a violin plot

FeatureScatter is typically used to visualize feature-feature relationships, but can be used

for anything calculated by the object, i.e. columns in object metadata, PC scores etc.

Normalizing the data

Identification of highly variable features (feature selection)

Identify the 10 most highly variable genes

plot variable features with and without labels

Scaling the data

▶如果这一步骤太久了，可以这么加快？

▶应该怎么去除一些不需要的变量来源，就像在Seurat v2那样？

Perform linear dimensional reduction

Examine and visualize PCA results a few different ways

Determine the 'dimensionality' of the dataset

NOTE: This process can take a long time for big datasets, comment out for expediency. More

approximate techniques such as those implemented in ElbowPlot() can be used to reduce

computation time

Cluster the cells

Look at cluster IDs of the first 5 cells

Run non-linear dimensional reduction (UMAP/tSNE)

If you haven't installed UMAP, you can do so via reticulate::py_install(packages ='umap-learn')

note that you can set label = TRUE or use the LabelClusters function to help label

#individual clusters

Finding differentially expressed features (cluster biomarkers)

find all markers of cluster 1

find all markers distinguishing cluster 5 from clusters 0 and 3

find markers for every cluster compared to all remaining cells, report only the positive ones

A tibble: 18 x 7

Groups: cluster [9]

you can plot raw counts as well

Assigning cell type identity to clusters

你可能感兴趣的:(单细胞测序--Seurat包--官网经典流程)

note that you can set `label = TRUE` or use the LabelClusters function to help label