WGCNA

RNA项目的WGCNA: 基因模块是否在不同分组间有差异？关注一组共表达的基因（i.e.模块）而不是单个差异基因

R

BiocManager::install("WGCNA") 进行安装

教程：Network analysis with WGCNA，中文讲解

输入：表达矩阵（行=样本，列=基因），需要知晓样本的分组；一般需要>15样本，且建议输入全部基因、不要仅选择差异基因

数据: input_mat, meta_df

## wget ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE61nnn/GSE61333/suppl/GSE61333_ligule_count.txt.gz
## gunzip GSE61333_ligule_count.txt.gz
library(tidyverse)    
library(magrittr)  
library(DESeq2)
library(WGCNA)   

## Load matrix
load('input_mat.rdata') ## normalized input matrix, we can check rna-counts of each sample via violin_plots
load('meta_df.rdata')   ## group_info

########## WGCNA ##########

## 1. Select soft-thresholding powers for network building
## sft$fitIndices:   Power, SFT.R.sq, slope, truncated.R.sq, mean.k., median.k., max.k
## X-axis: Power   (=sft$fitIndices[, 1])
## Y-axis: mean.k.   or   -1*SFT.R.sq*slope (sft$powerEstimate decided upon this value)
powers = c(c(1:10), seq(from = 12, to = 16, by = 2))
sft = pickSoftThreshold(input_mat, powerVector = powers,verbose = 5)


## 2. Build a complete network
net <- blockwiseModules(input_mat,power = sft$powerEstimate,networkType = "signed")


## 3. Plot modules, ...
moduleColors <- labels2colors(net$colors)
plotDendroAndColors(
  net$dendrograms[[1]],
  moduleColors[net$blockGenes[[1]]],
  "Module colors",
  dendroLabels = FALSE,
  hang = 0.03,
  addGuide = TRUE,
  guideHang = 0.05
)


## 4. Genes in each modules:  gene_id, colors
module_df <- data.frame(
  gene_id = names(net$colors),
  colors = moduleColors  ## labels2colors(netwk$colors)
)


## 5. Module-trait(Group) relationship: based on PC1 of each samples in each module (the so-called eigengenes)
sample_num <- nrow(value)
value <- moduleEigengenes(input_mat, moduleColors)$eigengenes
trait <- matrix(rnorm(sample_num) , nrow =sample_num)
p_val <- cor(value, trait, use = "p")
p_val_adj <- corPvalueStudent(p_val, sample_num)
# > p_val_adj
#                        [,1]
# MEblack         0.261063268
# MEblue          0.375112835
# MEbrown         0.955115638
# MEcyan          0.571728398

• 其中，灰色区域表示无法聚类到模块中的基因，如果占比过大，说明总体共表达趋势不明显，需要调整预处理方法/过滤基因
• 如果出现Error in (new("standardGeneric", .Data = function (x, y = NULL, use = "everything"，可能是包的冲突，强行令 cor <- WGCNA::cor