1 Description

PiGx scRNAseq

This report was generated with PiGx-scRNAseq version 0.0.1.

2 Input Settings

sceRdsFile = params$sceRdsFile
workdir = params$workdir
outFile = params$outFile
covariates = gsub(' ', '', unlist(strsplit(x = params$covariates, split = ',')))

inputParameterDesc = c('RDS format file containing a SingleCellExperiment object',
                     'Working directory',
                     'Path to HTML report output',
                     'Covariates to use when plotting (PCA and t-SNE)'
                     )
inputParameterValues = c(sceRdsFile,
                          workdir,
                          outFile,
                          paste0(covariates, collapse = ', '))
inputSettings = data.frame(parameters = inputParameterDesc,
                            values = inputParameterValues,
                            stringsAsFactors = FALSE)
DT::datatable(data = inputSettings,
              extensions = 'FixedColumns',
              options = list(fixedColumns = TRUE,
                         scrollX = TRUE,
                         pageLength = length(inputParameterValues),
                         dom = 't'))

2.1 Content Summary of the SingleCellExperiment Object

First a pre-generated SingleCellExperiment object is read.

This object must minimally have the following assays:

  • cnts (row read counts matrix per gene X cell)
  • cpm (cnts object normalized into counts per million in log2 scale)
  • scale (centered and scaled version of cpm)

And the following reduced dimension datasets:

  • PCA: Principle Component Analysis values for each cell (at least top 10 components)
  • TSNE: t-SNE analysis values of for each cell (at least top 2 components)
sce = readRDS(file = sceRdsFile)
print(sce)
## class: SingleCellExperiment 
## dim: 46603 31471 
## metadata(0):
## assays(3): cnts cpm scale
## rownames: NULL
## rowData names(8): Genes ndetected ... row_variability geneName
## colnames: NULL
## colData names(12): sample_name cell_id ... mean_expr
##   CellCyclePhase
## reducedDimNames(2): PCA TSNE
## spikeNames(0):

3 Results

plotCellStats = function(df, x, y, label_x, label_y, title) {
  p = ggplot(df, aes_string(x=x, y=y)) +
  geom_boxplot() +
  labs(x = label_x, y = label_y, title = title) +
  theme(plot.title = element_text(hjust = 0.5),
        axis.text.x = element_text(angle = 90, hjust = 1))
  return(p)
}

plotDesc = list('nGene' = c('Sample',
                          'Number of detected genes',
                          'Number of detected genes per cell'),
              'max_gene' = c('Sample',
                             'Maximum gene expression per cell',
                             'Maximum gene expression'),
              'mean_expr' = c('Sample',
                              'Average gene expression',
                              'Average gene expression per cell\nfor genes with >0 UMI'))

cellStatsPlots = lapply(names(plotDesc), function(y){
  p = plotCellStats(df = as.data.frame(colData(sce)),
                x = 'sample_name',
                y = y,
                label_x = plotDesc[[y]][1],
                label_y = plotDesc[[y]][2],
                title = plotDesc[[y]][3])
  return(p)
})
names(cellStatsPlots) = paste(lapply(plotDesc, function(x) x[2]))

3.1 Sample Statistics

3.1.1 Total number of reads per sample

The plot displays the numbers of mapped reads (in millions) for total (green bars) and uniquely-mapped UMI (orange bars) for each sample.

3.1.2 Read Distribution in Genomic Features

The plot displays the numbers of mapped reads (in millions) categorized by annotation (as indicated by the legend) for each sample.

3.2 Cell Statistics

These are boxplots of cell statistics.

Number of detected genes displays the distributions of numbers of detected genes per cell. Maximum gene expression per cell displays the distributions of maximum gene expression per cell. Average gene expression displays the distributions of average gene expression for genes with at least 1 UMI per cell.

3.2.1 Number of detected genes

3.2.2 Maximum gene expression per cell

3.2.3 Average gene expression

3.3 Cell Composition

The following plots visualize the cells in the principle component space, for the pairs of components indicated in the axes. These plots can be used to assess the data structure as well as to detect clusters, trends, outliers, and batch effects. The sample points are colored by the feature indicated in the corresponding tab title.

reducedDimPlot = function(df, dim1, dim2, title = NULL, color_by = NULL, gradient = FALSE) {
  p = ggplot2::ggplot(df, aes_string(x = dim1, y = dim2)) +
    geom_point(aes_string(color = color_by), size = 0.5, alpha = 0.5) +
    labs(title = title)

  if (gradient == TRUE) {
    p = p + scale_color_gradient2(midpoint = 3.3)
  }
  return(p)
}
colData(sce)$log10nGene = log10(colData(sce)$nGene)
gradientList = c('log10nGene')
covariateList = c(covariates, gradientList)
if('CellCyclePhase' %in% colnames(colData(sce))) {
  covariateList = c(covariateList, 'CellCyclePhase')  
}

pcaData = as.data.table(cbind(colData(sce), sce@reducedDims$PCA))

pcaPlotList <- lapply(covariateList, function(cov) {
  pL<- lapply(paste0('PC', c(2:5)), function(pc) {
  myTitle <- paste('PC1 vs', pc)
  reducedDimPlot(df = pcaData,
                 dim1 = 'PC1',
                 dim2 = pc,
                 title = myTitle,
                 color_by = cov,
                 gradient = cov %in% gradientList)
  })
names(pL) <- paste0('PC1 vs PC',c(2:5))

  return(pL)
  }
)
names(pcaPlotList) = covariateList

3.3.1 PCA - Colored by replicate

3.3.1.1 PC1 vs PC2

3.3.1.2 PC1 vs PC3

3.3.1.3 PC1 vs PC4

3.3.1.4 PC1 vs PC5

3.3.2 PCA - Colored by log10nGene

3.3.2.1 PC1 vs PC2

3.3.2.2 PC1 vs PC3

3.3.2.3 PC1 vs PC4