MVA_CNA: CopyNumbers_from_10xRNAseq using moving averages

Description

This is a straightforward method written in R to calculate copy numbers from 10X RNAseq data using moving averages of gene expression. Genome cordinates are annotated with hg20 and ensembl 93 through BioMart. Top genes that are expressed in 40% of either testing group of control group are selected to smoothing. The sliding windows are 50 genes. This method is designed to separate aneuploid tumor cells from diploid normal cells within tumor tissues. By default it will not help if you meet a diploid tumor.

Get ready

install {BioMart} using BiocInstaller::biocLite('grimbough/biomaRt')

To run

source("/volumes/lab/users/ruligao/code/update_MVA_function_version.R")

CNA_result <- cal_CNAs(tumor_mat, normal_mat, ROW.name="ENSEMBLE_id", plot=TRUE)

To prepare input

1. tumor_mat, is the UMI count matrix from testing sample, genes in the rows and cells in the columns
2. normal_mat, is the UMI count matrix from normal control sample, genes in the rows and cells in the columns
3. ROW.name, rownames of the input matrix, default "ENSEMBLE_id" or "GENE_SYMBOL"
4. plot=TRUE, default, will generate a heatmap of the copy numbers of all cells

Description of output

1. it outputs a list of two matrix objects
2. "MVA_results", the MVA results for all cells
3. "defined_celltype", defined tumor or normal cells by clustering
4. heatmap plot in PDF in working directory. always double check with the plot to make sure tumor cells have CNAs

An example run directly from 10X output

source("/volumes/lab/users/ruligao/code/update_MVA_function_version.R")

library(cellrangerRkit)

prepare count matrix of testing sample

path1 <- "/Volumes/seq/projects/ATC_thyroid/LAI_p259T" ##path that is one layer above outs/ from the 10x output

tumor_obj <- load_cellranger_matrix(path1)

tumor_mat <- exprs(tumor_obj)

prepare count matrix of normal control sample

path2 <- "/Volumes/seq/projects/ATC_thyroid/LAI_p259N"

normal_obj <- load_cellranger_matrix(path2)

normal_mat <- exprs(normal_obj)

CNA_result <- cal_CNAs(tumor_mat, normal_mat, ROW.name="ENSEMBLE_id", plot=TRUE)

An example run directly from Seurate object

source("/volumes/lab/users/ruligao/code/update_MVA_function_version.R")

tumor_mat <- as.matrix(tumor_obj@raw.data)

normal_mat <- as.matrix(normal_obj@raw.data)

CNA_result <- cal_CNAs(tumor_mat, normal_mat, ROW.name="GENE_SYMBOL", plot=TRUE)

An example run directly from count matrix

source("/volumes/lab/users/ruligao/code/update_MVA_function_version.R")

CNA_result <- cal_CNAs(tumor_mat, normal_mat, ROW.name="GENE_SYMBOL", plot=TRUE)

####### Updated by Ruli Gao, August 22nd, 2018