This tutorial show how to run InferCNV to detect CNVs in Cluster6 of snRNA-seq of the ROSMAP cohort.
The original InferCNV Tutorial: https://github.com/broadinstitute/inferCNV/wiki/Running-InferCNV
Before running codes below:
module load jags
R version: R/4.0.2

Load necessary packages and functions into R

rm(list = ls())
library(infercnv)
library(Seurat)
library(dplyr)
## 
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
## 
##     filter, lag
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union
outdir <- "CNV_results"
if(!file.exists(outdir)) dir.create(outdir, recursive = TRUE)

data loading

Now, since the data files is huge, load it from cloud. A precomputed version of the SeuratObject file is available at https://www.synapse.org/#!Synapse:syn26477245 Please download the precomputed SeuratObject first.

seurat_obj=readRDS("SeuratObject.RDS")
print(seurat_obj)
## An object of class Seurat 
## 35350 features across 70634 samples within 2 assays 
## Active assay: SCT (17575 features, 3000 variable features)
##  1 other assay present: RNA
##  3 dimensional reductions calculated: pca, umap, tsne

Subset Cluster 6:

Here, we extract cells of Cluster 6

mycluster=subset(x = seurat_obj, idents = 6)

Preparation before creating the infercnv object

Raw count matrix:

Next, we need to prepare several variables before constructing the InferCNV object Extract raw count matrix:

raw_counts_matrix=mycluster@assays$RNA@counts

Cell annotation file:

Prepare cell annotation file:

cellnames=row.names(mycluster@meta.data)
cellclusters=paste(mycluster@meta.data$SeuratCelltype,mycluster@meta.data$pathology.group,sep="_")
mycellAnnotation=t(t(cellclusters))
rownames(mycellAnnotation)=cellnames

Construct InferCNV object

Now we construct the inferCNV object
The “my_gen_pos.GRCh38.txt” is prepared gene location file of GRCh38.

infercnv_obj = CreateInfercnvObject(raw_counts_matrix=raw_counts_matrix,annotations_file=mycellAnnotation,delim="\t",gene_order_file="my_gen_pos.GRCh38.txt",ref_group_names=unique(cellclusters)[grep("no-pathology",unique(cellclusters))])

CNV detection

Now we run InferCNV to detect CNV:

infercnv_obj = infercnv::run(infercnv_obj,cutoff=0.1, out_dir=outdir cluster_by_groups=TRUE, denoise=TRUE, HMM=TRUE)

The CNV regions are stored in:HMM_CNV_predictions.HMMi6.hmm_mode-samples.Pnorm_0.5.pred_cnv_regions.dat and HMM_CNV_predictions.HMMi6.hmm_mode-samples.Pnorm_0.5.pred_cnv_genes.dat