Skip to contents

Compute log R ratios for scRNA-seq data

Source: R/computeLogRatio.R
computeLogRatioRNA.Rd

Compute log R ratios from raw count matrices with method specifically adapted for scRNA-seq data.

Usage

computeLogRatioRNA(
  matTumor,
  matRef,
  genesCoord,
  genome = "hg38",
  genesPerWindow = 101,
  refReads = 100,
  refMeanReads = 0.01,
  thresh_capping = 3,
  all_steps = FALSE,
  quiet = FALSE
)

Arguments

matTumor

Raw count matrix features x cells for tumor/sample cells (matrix or dgCMatrix).

matRef

Raw count matrix features x cells for reference cells (matrix or dgCMatrix).

genesCoord

Data frame of gene coordinates with columns CHROM, start, end, id (data.frame).

genome

Reference genome name among: "hg38", "hg19" and "mm10" (character). By default: "hg38".

genesPerWindow

Number of genes per moving window (integer value). By default: 101.

refReads

Minimum of reads in reference cells (integer value). By default: 100.

refMeanReads

Minimum of average reads per reference cell (integer value). By default: 0.01.

thresh_capping

Threshold to cap the range of log R ratio values (numeric value). By default: 3.

all_steps

TRUE or FALSE (logical). Whether to keep intermediate result from every step in the final object. By default: FALSE.

quiet

Logical. If TRUE, suppresses informative messages during execution. Default is FALSE.

Value

If all_steps is set to FALSE, a list containing:

matTumor

Matrix of log R ratio values features x cells for the tumor/sample cells (matrix).

matRef

Matrix of log R ratio values features x cells for the reference cells (matrix).

params

List of parameters set for the genesPerWindow, refMeans, refMeanReads and thresh_capping arguments (list).

coord

Data frame of coordinates for windows of peaks and associated data along the different steps (data.frame). Columns :

  • CHROM, start, end, id: coordinates and name of genes.

  • sumReads.tum/ref: sum of read counts for all cells in tumor or reference cells.

  • meanReads.tum/ref: mean of read counts per cells for tumor or reference cells.

  • sdReads.tum/ref: standard deviation of read counts per cells for tumor or reference cells.

  • keep: logical, TRUE for genes to keep after filtering based on reference coverage (depends on refReads and refMeanReads arguments).

  • meanReads/sdReads.norm.tum/ref: mean/sd of normalized counts per million for tumor/reference cells.

  • meanLRR/sdReads.raw.tum/ref: mean/sd of raw log R ratio (LRR) for tumor/reference cells.

  • meanLRR/sdLRR.cap.tum/ref: mean/sd of capped log R ratio (LRR) for tumor/reference cells (depends on thresh_capping argument).

  • meanLRR/sdLRR.smoo.tum/ref: mean/sd of smoothed log R ratio (LRR) for tumor/reference cells (means of moving windows defined by the genesPerWindow argument).

  • meanLRR/sdLRR.cent.tum/ref: mean/sd of centered log R ratio (LRR) for tumor/reference cells.

  • meanLRR/sdLRR.corr.tum/ref: mean/sd of final log R ratio (LRR) corrected by reference variability for tumor/reference cells.

If all_steps is set to TRUE, the previously described list can be found for each step in a list element.

Details

The raw count matrix is transformed into log R ratios through the following steps:

  • Match genes in count matrix with coordinates

  • Filtering on coverage (refReads and refMeanReads arguments)

  • Normalization for sequencing depth

  • Log transformation and normalization by reference data: log R ratio

  • Capping the range of values (thresh_capping argument)

  • Smoothing on genes windows

  • Centering of cells

  • Correcting by reference variability

See also

Other computeLogRatio: computeLogRatio(), computeLogRatioATAC()

Examples

# Create muscomic objects
atac <- CreateMuscomicObject(
  type = "ATAC",
  mat_counts = mat_counts_atac_tumor,
  allele_counts = allele_counts_atac_tumor,
  features = peaks
)
rna <- CreateMuscomicObject(
  type = "RNA",
  mat_counts = mat_counts_rna_tumor,
  allele_counts = allele_counts_rna_tumor,
  features = genes
)
atac_ref <- CreateMuscomicObject(
  type = "ATAC",
  mat_counts = mat_counts_atac_ref,
  allele_counts = allele_counts_atac_ref,
  features = peaks
)
rna_ref <- CreateMuscomicObject(
  type = "RNA",
  mat_counts = mat_counts_rna_ref,
  allele_counts = allele_counts_rna_ref,
  features = genes
)

# Create muscadet objects
muscadet <- CreateMuscadetObject(
  omics = list(atac, rna),
  bulk.lrr = bulk_lrr,
  bulk.label = "WGS",
  genome = "hg38"
)
muscadet_ref <- CreateMuscadetObject(
  omics = list(atac_ref, rna_ref),
  genome = "hg38"
)

# Compute log R ratio for RNA
obj_rna <- computeLogRatioRNA(
  matTumor = matCounts(muscadet)$RNA,
  matRef = matCounts(muscadet_ref)$RNA,
  genesCoord = coordFeatures(muscadet)$RNA,
  genome = slot(muscadet, "genome"),
  refReads = 2 # low value for example subsampled datasets
)
#> Step 01 - Match genes in count matrix with coordinates
#> Step 02 - Filtering genes: Minimum of 2 read(s) in reference cells and minimum of 0.01 read(s) in average per reference cell
#> Step 03 - Normalization for sequencing depth: Normalized counts per million
#> Step 04 - Log transformation and normalization by reference data: log R ratio
#> Step 05 - Capping the range of values: threshold = 3
#> Step 06 - Smoothing values on gene windows: 101 genes per window
#> Step 07 - Centering of cells
#> Step 08 - Correcting by reference variability
table(obj_rna$coord$keep)
#> 
#> FALSE  TRUE 
#>   151   349 

# With results form every step when `all_steps = TRUE`
obj_rna_all <- computeLogRatioRNA(
  matTumor = matCounts(muscadet)$RNA,
  matRef = matCounts(muscadet_ref)$RNA,
  genesCoord = coordFeatures(muscadet)$RNA,
  genome = slot(muscadet, "genome"),
  refReads = 2, # low value for example subsampled datasets
  all_steps = TRUE
)
#> Step 01 - Match genes in count matrix with coordinates
#> Step 02 - Filtering genes: Minimum of 2 read(s) in reference cells and minimum of 0.01 read(s) in average per reference cell
#> Step 03 - Normalization for sequencing depth: Normalized counts per million
#> Step 04 - Log transformation and normalization by reference data: log R ratio
#> Step 05 - Capping the range of values: threshold = 3
#> Step 06 - Smoothing values on gene windows: 101 genes per window
#> Step 07 - Centering of cells
#> Step 08 - Correcting by reference variability
names(obj_rna_all)
#>  [1] "step01" "step02" "step03" "step04" "step05" "step06" "step07" "step08"
#>  [9] "params" "coord" 
table(obj_rna_all$coord$keep)
#> 
#> FALSE  TRUE 
#>   151   349 
nrow(obj_rna_all$step08$matTumor)
#> [1] 349