【cellchat】Inference and analysis of cell-cell communication-基础篇，一个condition

时间：2024-10-07 07:29:07

cellchat需要两个输入：基因表达矩阵 + cell label。表达矩阵行为基因列为细胞，标准化+log。cell label为dataframe,行名为细胞，第一列为cell label.

从seurat v3 object中提取cell chat 所需要的两个输入文件：

    

  1. 

    

     <- GetAssayData(seurat_object, assay = "RNA", slot = "data") # normalized data matrix

    

    2. 

  2. 

    

    labels <- Idents(seurat_object)

    

    3. 

  3. 

    

    meta <- (group = labels,  = names(labels)) # create a dataframe of the cell labels

    

    4. 

    

  1. 

    

    library(CellChat)

    

    2. 

  2. 

    

    library(patchwork)

    

    3. 

  3. 

    

    options(stringsAsFactors = FALSE)

    

    4.

 

  • part I: data input and preprocessing, initialization of the cellchat object 

（1）load data

    

  1. 

    

    # Here we load a scRNA-seq data matrix and its associated cell meta data
    

    2. 

  2. 

    

    #load(url("/files/25950872")) 
    

    3. 

  3. 

    

    # This is a combined data from two biological conditions: normal and diseases
    

    4. 

  4. 

    

     
    

    5. 

  5. 

    

    load("/Users/suoqinjin/Documents/CellChat/tutorial/data_humanSkin_CellChat.rda")

    

    6. 

  6. 

    

     
    

    7. 

  7. 

    

     = data_humanSkin$data # normalized data matrix

    

    8. 

  8. 

    

     
    

    9. 

  9. 

    

    meta = data_humanSkin$meta # a dataframe with rownames containing cell mata data

    

    10. 

  10. 

    

     
    

    11. 

  11. 

    

     = rownames(meta)[meta$condition == "LS"] # extract the cell names from disease data

    

    12. 

  12. 

    

     
    

    13. 

  13. 

    

    # Prepare input data for CelChat analysis
    

    14. 

  14. 

    

     = [, ]

    

    15. 

  15. 

    

    meta = meta[, ]

    

    16. 

  16. 

    

    # meta = (labels = meta$labels[],  = colnames()) # manually create a dataframe consisting of the cell labels
    

    17. 

  17. 

    

     
    

    18. 

  18. 

    

    unique(meta$labels) # check the cell labels

    

    19. 

  19. 

    

    #>  [1] Inflam. FIB  FBN1+ FIB    APOE+ FIB    COL11A1+ FIB cDC2        
    

    20. 

  20. 

    

    #>  [6] LC           Inflam. DC   cDC1         CD40LG+ TC   Inflam. TC  
    

    21. 

  21. 

    

    #> [11] TC           NKT         
    

    22. 

  22. 

    

    #> 12 Levels: APOE+ FIB FBN1+ FIB COL11A1+ FIB Inflam. FIB cDC1 cDC2 ... NKT
    

    23.

 （2）create a cellchat object

cellchat <- createCellChat(object = , meta = meta,  = "labels")

（3）set the ligand-receptor interaction database

    

  1. 

    

    CellChatDB <-  # use  if running on mouse data

    

    2. 

  2. 

    

    showDatabaseCategory(CellChatDB)

    

    3. 

  3. 

    

     
    

    4. 

  4. 

    

    # Show the structure of the database
    

    5. 

  5. 

    

    dplyr::glimpse(CellChatDB$interaction)

    

    6. 

  6. 

    

     
    

    7. 

  7. 

    

    # use a subset of CellChatDB for cell-cell communication analysis
    

    8. 

  8. 

    

     <- subsetDB(CellChatDB, search = "Secreted Signaling") # use Secreted Signaling

    

    9. 

  9. 

    

     
    

    10. 

  10. 

    

    # use all CellChatDB for cell-cell communication analysis
    

    11. 

  11. 

    

    #  <- CellChatDB # simply use the default CellChatDB
    

    12. 

  12. 

    

     
    

    13. 

  13. 

    

    # set the used database in the object
    

    14. 

  14. 

    

    cellchat@DB <- 
    

    15.

 （4）preprocessing the gene expression data

    

  1. 

    

    # subset the expression data of signaling genes for saving computation cost
    

    2. 

  2. 

    

    cellchat <- subsetData(cellchat) # This step is necessary even if using the whole database

    

    3. 

  3. 

    

    future::plan("multiprocess", workers = 4) # do parallel

    

    4. 

  4. 

    

     
    

    5. 

  5. 

    

    cellchat <- identifyOverExpressedGenes(cellchat)

    

    6. 

  6. 

    

    cellchat <- identifyOverExpressedInteractions(cellchat)

    

    7. 

  7. 

    

    # project gene expression data onto PPI network (optional)
    

    8. 

  8. 

    

    cellchat <- projectData(cellchat, )

    

    9.

 

  •  Part II: Inference of cell-cell communication network

在核心功能computeCommuProb中，可以设置不同的阈值。默认为trimean(25%),也可以设置成10%、5%等。这个阈值什么意思？当某个cell group中表达某个pair的细胞比例低于此值时，他们的平均表达量就都是零（不会出现在结果里），调低阈值能发现更多通讯，但相应的噪音也更大。调整方法：type = "truncatedMean", trim = 0.1

（1）compute the communication probability and infer the celluar communication network 

    

  1. 

    

    cellchat <- computeCommunProb(cellchat)

    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    # Filter out the cell-cell communication if there are only few number of cells in certain cell groups
    

    4. 

  4. 

    

    cellchat <- filterCommunication(cellchat,  = 10)

    

    5.

（2） extract the infered celluar communication network as a dataframe

We provide a function subsetCommunication to easily access the inferred cell-cell communications of interest. For example,

  • <- subsetCommunication(cellchat) returns a data frame consisting of all the inferred cell-cell communications at the level of ligands/receptors. Set  = "netP" to access the the inferred communications at the level of signaling pathways

  • <- subsetCommunication(cellchat, = c(1,2), = c(4,5)) gives the inferred cell-cell communications sending from cell groups 1 and 2 to cell groups 4 and 5.

  • <- subsetCommunication(cellchat, signaling = c("WNT", "TGFb")) gives the inferred cell-cell communications mediated by signaling WNT and TGFb.

（3）infer the communication network at a signaling pathway level

注：前面以pair为单位的结果存储在“net” slot, 而pathway推断结果在“netP”slot.

cellchat <- computeCommunProbPathway(cellchat)

（4）calculate the aggregated communication network

USER can also calculate the aggregated network among a subset of cell groups by setting  and .

cellchat <- aggregateNet(cellchat)

两种可视化思路：一种线条粗细反应数量，一种反映weights.

    

  1. 

    

    groupSize <- (table(cellchat@idents))

    

    2. 

  2. 

    

    par(mfrow = c(1,2), xpd=TRUE)

    

    3. 

  3. 

    

     
    

    4. 

  4. 

    

    netVisual_circle(cellchat@net$count,  = groupSize,  = T, = F,  = "Number of interactions")

    

    5. 

  5. 

    

     
    

    6. 

  6. 

    

    netVisual_circle(cellchat@net$weight,  = groupSize,  = T, = F,  = "Interaction weights/strength")

    

    7.

 可视化从每种细胞类型发出的

    

  1. 

    

    mat <- cellchat@net$weight
    

    2. 

  2. 

    

    par(mfrow = c(3,4), xpd=TRUE)

    

    3. 

  3. 

    

    
for (i in 1:nrow(mat)) {

    

    4. 

  4. 

    

      mat2 <- matrix(0, nrow = nrow(mat), ncol = ncol(mat), dimnames = dimnames(mat))

    

    5. 

  5. 

    

      mat2[i, ] <- mat[i, ]

    

    6. 

  6. 

    

      netVisual_circle(mat2,  = groupSize,  = T,  = max(mat),  = rownames(mat)[i])

    

    7. 

  7. 

    

    }
    

    8.

 

  •  part III : visualization of the communication network

（1）visualize each signaling pathway 

所有显著的通讯通路都可以通过cellchat@netP$pathways查看

    

  1. 

    

     <- c("CXCL") 
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    # Hierarchy plot
    

    4. 

  4. 

    

    # Here we define `` so that the left portion of the hierarchy plot shows signaling to fibroblast and the right portion shows signaling to immune cells 
    

    5. 

  5. 

    

     
    

    6. 

  6. 

    

     = seq(1,4) # a numeric vector. 

    

    7. 

  7. 

    

    netVisual_aggregate(cellchat, signaling = ,   = )

    

    8.

 

    

  1. 

    

    # Circle plot
    

    2. 

  2. 

    

    par(mfrow=c(1,1))

    

    3. 

  3. 

    

    netVisual_aggregate(cellchat, signaling = , layout = "circle")

    

    4. 


    

  1. 

    

    # Chord diagram
    

    2. 

  2. 

    

    par(mfrow=c(1,1))

    

    3. 

  3. 

    

    netVisual_aggregate(cellchat, signaling = , layout = "chord")

    

    4. 

  4. 

    

    #> Note: The first link end is drawn out of sector 'Inflam. FIB'.
    

    5. 


    

  1. 

    

    # Heatmap
    

    2. 

  2. 

    

    par(mfrow=c(1,1))

    

    3. 

  3. 

    

    netVisual_heatmap(cellchat, signaling = ,  = "Reds")

    

    4.

 

 

（2）Compute the contribution of each ligand-receptor pair to the overall signaling pathway and visualize cell-cell communication mediated by a single ligand-receptor pair

netAnalysis_contribution(cellchat, signaling = )

We provide a function extractEnrichedLR to extract all the significant interactions (L-R pairs) and related signaling genes for a given signaling pathway.

    

  1. 

    

     <- extractEnrichedLR(cellchat, signaling = ,  = FALSE)

    

    2. 

  2. 

    

     <- [1,] # show one ligand-receptor pair

    

    3. 

  3. 

    

     
    

    4. 

  4. 

    

    # Hierarchy plot
    

    5. 

  5. 

    

     = seq(1,4) # a numeric vector

    

    6. 

  6. 

    

    netVisual_individual(cellchat, signaling = ,   = ,  = )

    

    7. 


    

  1. 

    

    # Circle plot
    

    2. 

  2. 

    

    netVisual_individual(cellchat, signaling = ,  = , layout = "circle")

    

    3. 


    

  1. 

    

    # Chord diagram
    

    2. 

  2. 

    

    netVisual_individual(cellchat, signaling = ,  = , layout = "chord")

    

    3.

 （3）visualize cell communications by multiple L-R pairs or pathways

气泡图

    

  1. 

    

    # show all the significant interactions (L-R pairs) from some cell groups (defined by '') to other cell groups (defined by '')
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    netVisual_bubble(cellchat,  = 4,  = c(5:11),  = FALSE)

    

    4.

 

    

  1. 

    

    # show all the significant interactions (L-R pairs) associated with certain signaling pathways
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    netVisual_bubble(cellchat,  = 4,  = c(5:11), signaling = c("CCL","CXCL"),  = FALSE)

    

    4. 


    

  1. 

    

    # show all the significant interactions (L-R pairs) based on user's input (defined by ``)
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

     <- extractEnrichedLR(cellchat, signaling = c("CCL","CXCL","FGF"))

    

    4. 

  4. 

    

    netVisual_bubble(cellchat,  = c(3,4),  = c(5:8),  = ,  = TRUE)

    

    5.

 和弦图（和bubble图一样，既可以展示用户指定的target与sender之间的通讯关系，也可以展示用户指定的pair和pathway）

    

  1. 

    

    # show all the significant interactions (L-R pairs) from some cell groups (defined by '') to other cell groups (defined by '')
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    # show all the interactions sending from 
    

    4. 

  4. 

    

    netVisual_chord_gene(cellchat,  = 4,  = c(5:11),  = 0.5, = 30)

    

    5. 


    

  1. 

    

    # show all the interactions received by 
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    netVisual_chord_gene(cellchat,  = c(1,2,3,4),  = 8,  = 15)

    

    4. 


    

  1. 

    

    # show all the significant interactions (L-R pairs) associated with certain signaling pathways
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    netVisual_chord_gene(cellchat,  = c(1,2,3,4),  = c(5:11), signaling = c("CCL","CXCL"), = 8)

    

    4. 


    

  1. 

    

    # show all the significant signaling pathways from some cell groups (defined by '') to other cell groups (defined by '')
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    netVisual_chord_gene(cellchat,  = c(1,2,3,4),  = c(5:11),  = "netP",  = 10)

    

    4.

 （4）plot the signaling gene expression distrobution by dot or violin plots

    

  1. 

    

    #来自seurat的wrapper功能
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    plotGeneExpression(cellchat, signaling = "CXCL")

    

    4.

 

  • part IV : systems analysis of cell-cell communication network

这是什么意思呢？

a. 根据已经构建好的通讯网络，发现其中的major sources, targets, influencers, mediators.

b. 对某些细胞类型，预测其关键的incoming and outgoing signals，包括coordinated responses among different cell types

c. group signaling pathways

d. delineate conserved and context-specific signaling pathways

（1）identify signaling roles of cell groups as well as the major contributing signaling (dominant senders and receivers)

    

  1. 

    

    # Compute the network centrality scores
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    cellchat <- netAnalysis_computeCentrality(cellchat,  = "netP") 
    

    4. 

  4. 

    

    # the slot 'netP' means the inferred intercellular communication network of signaling pathways
    

    5. 

  5. 

    

     
    

    6. 

  6. 

    

     
    

    7. 

  7. 

    

    # Visualize the computed centrality scores using heatmap, allowing ready identification of major signaling roles of cell groups
    

    8. 

  8. 

    

     
    

    9. 

  9. 

    

    netAnalysis_signalingRole_network(cellchat, signaling = , width = 8, height = 2.5,  = 10)

    

    10.

 Visualize the dominant senders (sources) and receivers (targets) in a 2D space

    

  1. 

    

    # Signaling role analysis on the aggregated cell-cell communication network from all signaling pathways
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    gg1 <- netAnalysis_signalingRole_scatter(cellchat)

    

    4. 

  4. 

    

     
    

    5. 

  5. 

    

    # Signaling role analysis on the cell-cell communication networks of interest
    

    6. 

  6. 

    

     
    

    7. 

  7. 

    

    gg2 <- netAnalysis_signalingRole_scatter(cellchat, signaling = c("CXCL", "CCL"))

    

    8. 

  8. 

    

     
    

    9. 

  9. 

    

    gg1 + gg2
    

    10.

 identify signals contributing most to the incoming and outgoing signalings of certain cell types

    

  1. 

    

    # Signaling role analysis on the aggregated cell-cell communication network from all signaling pathways
    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    ht1 <- netAnalysis_signalingRole_heatmap(cellchat, pattern = "outgoing")

    

    4. 

  4. 

    

    ht2 <- netAnalysis_signalingRole_heatmap(cellchat, pattern = "incoming")

    

    5. 

  5. 

    

    ht1 + ht2
    

    6. 


    

  1. 

    

    # Signaling role analysis on the cell-cell communication networks of interest
    

    2. 

  2. 

    

    ht <- netAnalysis_signalingRole_heatmap(cellchat, signaling = c("CXCL", "CCL"))

    

    3.

 

（2）identify global communication patterns to explore how multiple cell types and signaling pathways coordinate together

default patterns number: 5；可以用select K功能寻找最佳K值，选当曲线开始突然下降时的K。

identify outgoing communication patterns of secreting cells

    

  1. 

    

    library(NMF)

    

    2. 

  2. 

    

    library(ggalluvial)

    

    3. 

  3. 

    

     
    

    4. 

  4. 

    

    selectK(cellchat, pattern = "outgoing")

    

    5.

 k选择3

    

  1. 

    

    nPatterns = 3

    

    2. 

  2. 

    

     
    

    3. 

  3. 

    

    cellchat <- identifyCommunicationPatterns(cellchat, pattern = "outgoing", k = nPatterns)

    

    4.

 

    

  1. 

    

    # river plot
    

    2. 

  2. 

    

    netAnalysis_river(cellchat, pattern = "outgoing")

    

    3.

 

    

  1. 

    

    # dot plot
    

    2. 

  2. 

    

    netAnalysis_dot(cellchat, pattern = "outgoing")

    

    3.

 

identify incoming communication patterns of target cells功能都一样，把outgoing改成incoming就行

（3）manifold and classication learning analysis of signaling networks

可以对所有显著的signaling pathway进行grouping分析，依据通路之间的相似性，这种相似性既可以是功能上的“functional”,也可以是结构上的“structural”.

functional similarity: 主要的sender和receiver相同

structural similarity: 网络的结构相似，不考虑sender和receiver的相似性

identify signaling groups based on their functional similarity

    

  1. 

    

    cellchat <- computeNetSimilarity(cellchat, type = "functional")

    

    2. 

  2. 

    

    cellchat <- netEmbedding(cellchat, type = "functional")

    

    3. 

  3. 

    

     
    

    4. 

  4. 

    

     
    

    5. 

  5. 

    

    cellchat <- netClustering(cellchat, type = "functional")

    

    6. 

  6. 

    

     
    

    7. 

  7. 

    

    # Visualization in 2D-space
    

    8. 

  8. 

    

    netVisual_embedding(cellchat, type = "functional",  = 3.5)

    

    9. 

  9. 

    

     
    

    10. 

  10. 

    

    # netVisual_embeddingZoomIn(cellchat, type = "functional", nCol = 2)
    

    11.

 identify signaling groups based on structure similarity

    

  1. 

    

    cellchat <- computeNetSimilarity(cellchat, type = "structural")

    

    2. 

  2. 

    

    cellchat <- netEmbedding(cellchat, type = "structural")

    

    3. 

  3. 

    

     
    

    4. 

  4. 

    

    cellchat <- netClustering(cellchat, type = "structural")

    

    5. 

  5. 

    

     
    

    6. 

  6. 

    

    # Visualization in 2D-space
    

    7. 

  7. 

    

    netVisual_embedding(cellchat, type = "structural",  = 3.5)

    

    8. 


netVisual_embeddingZoomIn(cellchat, type = "structural", nCol = 2)

 

 

  • part V : save the cellchat object
saveRDS(cellchat, file = "cellchat_humanSkin_LS.rds")

这篇文章是关于单个dataset单个条件的,如果涉及到代表不同conditions的datasets 之间的差异分析，可以看另一个教程。

GitHub & BitBucket HTML Preview

相关文章