Title:	Geometric Single Cell Deconvolution
Version:	0.5.4
Description:	Deconvolution of bulk RNA-Sequencing data into proportions of cells based on a reference single-cell RNA-Sequencing dataset using high-dimensional geometric methodology.
License:	GPL (≥ 3)
Encoding:	UTF-8
Imports:	circlize, ComplexHeatmap, DelayedArray, dplyr, ensembldb, ggplot2, ggrepel, grid, gtools, matrixStats, mcprogress, parallel, pbmcapply, rlang, scales
RoxygenNote:	7.3.3
Depends:	R (≥ 4.1.0)
Suggests:	future.apply, ggsci, knitr, plotly, Rfast2, rmarkdown, seriation
VignetteBuilder:	knitr
NeedsCompilation:	no
Packaged:	2025-09-10 12:31:17 UTC; myles
Author:	Myles Lewis [aut, cre], Rachel Lau [ctb]
Maintainer:	Myles Lewis <myles.lewis@qmul.ac.uk>
Repository:	CRAN
Date/Publication:	2025-09-15 08:20:02 UTC

Add noise to count data

Description

Gaussian noise can be added to the simulated count matrix in multiple ways which can be combined.

Usage

add_noise(counts, sd = 100)

log_noise(counts, sd = 0.1)

graded_log_noise(counts, sd = 0.1, transform = function(x) x^3)

sqrt_noise(counts, sd = 100)

shift_noise(counts, sd = 0.5, p = 0.5)

Arguments

Details

• add_noise adds simple Gaussian noise to counts. This affects low expressed genes and hardly affects highly expressed genes.

• With log_noise, counts are converted using log2+1 and Gaussian noise added, followed by conversion back to count scale. This affects all genes irrespective of expression level.

• With graded_log_noise, counts are converted to log2+1. A scaling factor is calculated for gene expression level ranging from 0 to 1, which maps to 0 to the maximum number of counts. This scaling factor is inverted from 1 to 0 (i.e. noise affects low counts more than high counts) and then passed through the function specified by transform (this controls how much the middle counts are affected). Then the Gaussian noise is multiplied by the scaling factor and added to the counts.

• With sqrt_noise, counts are square root transformed before Gaussian noise is added, and then transformed back. This still has a stronger effect on low expressed genes, but the effect is more graduated with a more gradual fall off in effect on genes with increasing expression.

• With shift_noise, whole gene rows are selected at random then each row is multiplied by a random amount varying according to 2^rnorm. This simulates shifted expression up/down due to differences in chemistry through which some genes are more or less detectable.

Value

A positive integer count matrix with genes in rows and cell subclasses in columns.

Adjust count matrix by library size

Description

Simple tool for adjusting raw count matrix by total library size. Library size is calculated as column sums and columns are scaled to the median total library size.

Usage

adjust_library_size(x)

Arguments

Value

Matrix of adjusted read counts

Identify cell markers

Description

Uses geometric method based on vector dot product to identify genes which are the best markers for individual cell types.

Usage

cellMarkers(
  scdata,
  bulkdata = NULL,
  subclass,
  cellgroup = NULL,
  nsubclass = 25,
  ngroup = 10,
  expfilter = 0.5,
  noisefilter = 2,
  noisefraction = 0.25,
  min_cells = 10,
  remove_subclass = NULL,
  dual_mean = FALSE,
  meanFUN = "logmean",
  postFUN = NULL,
  verbose = TRUE,
  sliceMem = 16,
  cores = 1L,
  ...
)

Arguments

Details

If verbose = TRUE, the function will display an estimate of the required memory. But importantly this estimate is only a guide. It is provided to help users choose the optimal number of cores during parallelisation. Real memory usage might well be more, theoretically up to double this amount, due to R's use of copy-on-modify.

Value

A list object with S3 class 'cellMarkers' containing:

The 'cellMarkers' object is designed to be passed to deconvolute() to deconvolute bulk RNA-Seq data. It can be updated rapidly with different settings using updateMarkers(). Ensembl gene ids can be substituted for recognisable gene symbols by applying gene2symbol().

Author(s)

Myles Lewis

See Also

deconvolute() updateMarkers() gene2symbol()

Collapse groups in cellMarkers object

Description

Experimental function for collapsing groups in a cellMarkers objects.

Usage

collapse_group(mk, groups, weights = NULL)

Arguments

Value

An updated cellMarkers class object.

Compensation heatmap

Description

Plots a heatmap of the compensation matrix for cell subclasses using ComplexHeatmap.

Usage

comp_heatmap(
  x,
  cell_table = NULL,
  text = NULL,
  cutoff = 0.2,
  fontsize = 8,
  subset = NULL,
  ...
)

Arguments

Value

No return value. Draws a ComplexHeatmap.

Gene signature cosine similarity matrix

Description

Computes the cosine similarity matrix from the gene signature matrix of a cellMarkers object or any matrix. Note that this function computes cosine similarity between matrix columns, unlike dist() which computes the distance metric between matrix rows.

Usage

cos_similarity(x, use_filter = NULL)

Arguments

Value

A symmetric similarity matrix.

Deconvolute bulk RNA-Seq using single-cell RNA-Seq signature

Description

Deconvolution of bulk RNA-Seq using vector projection method with adjustable compensation for spillover.

Usage

deconvolute(
  mk,
  test,
  log = TRUE,
  count_space = TRUE,
  comp_amount = 1,
  group_comp_amount = 0,
  weights = NULL,
  weight_method = "equal",
  adjust_comp = TRUE,
  use_filter = TRUE,
  arith_mean = FALSE,
  convert_bulk = FALSE,
  check_comp = FALSE,
  npass = 1,
  outlier_method = c("var.e", "cooks", "rstudent"),
  outlier_cutoff = switch(outlier_method, var.e = 4, cooks = 1, rstudent = 10),
  outlier_quantile = 0.9,
  verbose = TRUE,
  cores = 1L
)

Arguments

Details

Equal weighting of genes by setting weight_method = "equal" can help devolution of subclusters whose signature genes have low expression. It is enabled by default.

Multipass deconvolution can be activated by setting npass to 2 or higher. This is designed to remove genes which behave inconsistently due to noise in either the sc or bulk datasets, which is increasingly likely if you have larger signature geneset, i.e. if nsubclass is large. Or you may receive a warning message "Detected genes with extreme residuals". Three methods are available for identifying outlier genes (i.e. whose residuals are too noisy) controlled by outlier_method:

• var.e, this calculates the variance of the residuals across samples for each gene. Genes whose variance of residuals are outliers based on Z-score standardisation are removed during successive passes.

• cooks, this considers the deconvolution as if it were a regression and applies Cook's distance to the residuals and the hat matrix. This seems to be the most stringent method (removes fewest genes).

• rstudent, externally Studentized residuals are used.

The cutoff specified by outlier_cutoff which is used to determine which genes are outliers is very sensitive to the outlier method. With var.e the variances are Z-score scaled. With Cook's distance it is typical to consider a value of >1 as fairly strong indication of an outlier, while 0.5 is considered a possible outlier. With Studentized residuals, these are expected to be on a t distribution scale. However, since gene expression itself does not derive from a normal distribution, the errors and residuals are not normally distributed either, which probably explains the need for a very high cut-off. In practice the choice of settings seems to be dataset dependent.

Value

A list object of S3 class 'deconv' containing:

Author(s)

Myles Lewis

See Also

cellMarkers() updateMarkers() rstudent.deconv() cooks.distance.deconv()

Diagnostics for cellMarker signatures

Description

Diagnostic tool which prints information for identifying cell subclasses or groups with weak signatures.

Usage

diagnose(object, group = NULL, angle_cutoff = 30, weak = 2)

Arguments

Value

No return value. Prints information about the cellMarkers signature showing cells subclasses with weak signatures and diagnostic information including which cell subclasses each problematic signature spills into.

Fix in missing genes in bulk RNA-Seq matrix

Description

Fills in missing genes in a bulk RNA-Seq matrix based on the gene signature of a 'cellMarkers' objects. Signature is taken from both the subclass gene set and group gene set.

Usage

fix_bulk(bulk, mk)

Arguments

Details

This is a convenience function if you have an existing cellMarkers signature object and you do not want to remove genes from the existing signatures by running updateMarkers() with the desired bulk data, and are prepared to accept the assumption that genes which are missing in the bulk RNA-Seq dataset have zero expression. We recommend you check which signature genes are missing from the bulk data first.

Value

Expanded bulk matrix with extra rows for missing genes, filled with zeros.

Converts ensembl gene ids to symbols

Description

Uses a loaded ensembl database to convert ensembl gene ids to symbol. If a vector is provided, a vector of symbols is returned. If a cellMarkers object is provided, the rownames in the genemeans, genemeans_filtered, groupmeans and groupmeans_filtered elements are changed to symbol and the cellMarkers object is returned.

Usage

gene2symbol(x, ensdb, dups = c("omit", "pass"))

Arguments

Value

If x is a vector, a vector of symbols is returned. If no symbol is available for particular ensembl id, the id is left untouched. If x is a 'cellMarkers' class object, a 'cellMarkers' object is returned with rownames in the results elements and genesets converted to gene symbols, and an extra element symbol containing a named vector of converted genes.

See Also

cellMarkers()

Vector based best marker selection

Description

Core function which takes a matrix of mean gene expression (assumed to be log2 transformed to be more Gaussian). Mean gene expression per gene is scaled to a unit hypersphere assuming each gene represents a vector in space with dimensions representing each cell subclass/group.

Usage

gene_angle(genemeans)

Arguments

Value

a list whose length is the number of columns in genemeans, with each element containing a dataframe with genes in rows, sorted by best marker status as determined by minimum vector angle and highest maximum gene expression per celltype/tissue.

Generate random cell number samples

Description

Used for simulating pseudo-bulk RNA-Seq from a 'cellMarkers' object. Cell counts are randomly sampled from the uniform distribution, using the original subclass contingency table as a limit on the maximum number of cells in each subclass.

Usage

generate_samples(
  object,
  n,
  equal_sample = TRUE,
  method = c("unif", "dirichlet"),
  alpha = 1.5
)

Arguments

Details

Leaving equal_sample = TRUE is better for tuning deconvolution parameters.

Value

An integer matrix with n rows, with columns for each cell subclasses in object, representing cell counts for each cell subclass. Designed to be passed to simulate_bulk().

See Also

simulate_bulk()

Mean Objects

Description

Functions designed for use with scmean() to calculate mean gene expression in each cell cluster across matrix rows.

Usage

logmean(x)

trimmean(x)

log2s(x)

Arguments

Value

Numeric vector of mean values.

logmean applies log2(x+1) then calculates rowMeans.

trimmean applies a trimmed mean to each row of gene counts, excluding the top and bottom 5% of values which helps to exclude outliers. Note, this needs the Rfast2 package to be installed. When trimmean is used with scmean(), postFUN is typically set to log2s. This simply applies log2(x+1) after the trimmed mean of counts has been calculated.

Merge cellMarker signatures

Description

Takes 2 cellMarkers signatures, merges them and recalculates optimal gene signatures.

Usage

mergeMarkers(
  mk1,
  mk2,
  remove_subclass = NULL,
  remove_group = NULL,
  transform = c("qq", "linear.qq", "scale", "none"),
  scale = 1,
  ...
)

Arguments

Value

A list object of S3 class 'cellMarkers'. See cellMarkers() for details. If transform = "qq" then an additional element qqmerge is returned containing the quantile mapping function between the 2 datasets.

See Also

cellMarkers() updateMarkers() quantile_map()

Calculate R-squared and metrics on deconvoluted cell subclasses

Description

Calculates Pearson r-squared, R-squared and RMSE comparing subclasses in each column of obs with matching columns in deconvoluted pred. Samples are in rows. For use if ground truth is available, e.g. simulated pseudo-bulk RNA-Seq data.

Usage

metric_set(obs, pred)

Arguments

Details

Pearson r-squared ranges from 0 to 1. R-squared, calculated as 1 - rss/tss, ranges from -Inf to 1.

Value

Matrix containing Pearson r-squared, R-squared and RMSE values.

Quantile-quantile plot

Description

Produces a QQ plot showing the conversion function from the first dataset to the second.

Usage

## S3 method for class 'qqmap'
plot(x, points = TRUE, ...)

Arguments

Value

No return value. Produces a QQ plot using base graphics with a red line showing the conversion function.

Plot compensation analysis

Description

Plots the effect of varying compensation from 0 to 1 for each cell subclass, examining the minimum subclass output result following a call to deconvolute(). For this function to work, the argument plot_comp must be set to TRUE during the call to deconvolute().

Usage

plot_comp(x, overlay = TRUE, mfrow = NULL, ...)

Arguments

Value

No return value, plots the effect of varying compensation on minimum subclass output for each cell subclass.

Residuals plot

Description

Plots residuals from a deconvolution result object against bulk gene expression (on semi-log axis). Normal residuals, weighted residuals or Studentized residuals can be visualised to check for heteroscedasticity and genes with extreme errors.

Usage

plot_residuals(
  fit,
  test,
  type = c("reg", "student", "weight"),
  show_outliers = TRUE,
  show_plot = TRUE,
  ...
)

ggplot_residuals(
  fit,
  test,
  type = c("reg", "student", "weight"),
  show_outliers = TRUE
)

Arguments

Value

Produces a scatter plot in base graphics. Returns invisibly a dataframe of the coordinates of the points. The ggplot version returns a ggplot2 plotting object.

Scatter plots to compare deconvoluted subclasses

Description

Produces scatter plots using base graphics to compare actual cell counts against deconvoluted cell counts from bulk (or pseudo-bulk) RNA-Seq. Mainly for use if ground truth is available, e.g. for simulated pseudo-bulk RNA-Seq data.

Usage

plot_set(
  obs,
  pred,
  mfrow = NULL,
  show_zero = FALSE,
  show_identity = FALSE,
  cols = NULL,
  colour = "blue",
  title = "",
  cex.title = 1,
  ...
)

Arguments

Value

No return value. Produces scatter plots using base graphics.

Plot tuning curves

Description

Produces a ggplot2 plot of R-squared/RMSE values generated by tune_deconv().

Usage

plot_tune(
  result,
  group = "subclass",
  xvar = colnames(result)[1],
  fix = NULL,
  metric = attr(result, "metric"),
  title = NULL
)

Arguments

Details

If group is set to "subclass", then the tuning parameter specified by xvar is varied on the x axis. Any other tuning parameters (i.e. if 2 or more have been tuned) are fixed to their best tuned values.

If group is set to a different column than "subclass", then the mean R-squared/RMSE values in result are averaged over subclasses. This makes it easier to compare the overall effect (mean R-squared/RMSE) of 2 tuned parameters which are specified by xvar and group. Any remaining parameters not shown are fixed to their best tuned values.

If group is NULL, the tuning parameter specified by xvar is varied on the x axis and R-squared/RMSE values are averaged over the whole grid to give the generalised mean effect of varying the xvar parameter.

Value

ggplot2 scatter plot.

Quantile mapping function between two scRNA-Seq datasets

Description

Quantile mapping to combine two scRNA-Seq datasets based on mapping either the distribution of mean log2+1 gene expression in cell clusters to the distribution of the 2nd dataset, or mapping the quantiles of one matrix of gene expression (with genes in rows) to another.

Usage

quantile_map(
  x,
  y,
  n = 10000,
  remove_noncoding = TRUE,
  remove_zeros = FALSE,
  smooth = "loess",
  span = 0.15,
  knots = c(0.25, 0.75, 0.85, 0.95, 0.97, 0.99, 0.999),
  respace = FALSE,
  silent = FALSE
)

Arguments

Details

The conversion uses the function approxfun() which uses interpolation. It is not designed to perform stepwise (exact) quantile transformation of every individual datapoint.

Value

A list object of class 'qqmap' containing:

See Also

approxfun()

Rank distance angles from a cosine similarity matrix

Description

Converts a cosine similarity matrix to angular distance. Then orders the elements in increasing angle. Elements below angle_cutoff are returned in a dataframe.

Usage

rank_angle(x, angle_cutoff = 45)

Arguments

Value

a dataframe of rows and columns as factors and the angle between that row and column extracted from the cosine similarity matrix. Row and column location are stored as factors so that they can be converted back to coordinates in the similarity matrix easily using as.integer().

Reduce noise in single-cell data

Description

Simple filter for removing noise in single-cell data.

Usage

reduceNoise(cellmat, noisefilter = 2, noisefraction = 0.25)

Arguments

Value

Filtered mean gene expression matrix with genes in rows and cell types in columns.

Regression Deletion Diagnostics

Description

Functions for computing regression diagnostics including standardised or Studentized residuals as well as Cook's distance.

Usage

## S3 method for class 'deconv'
rstudent(model, ...)

## S3 method for class 'deconv'
rstandard(model, ...)

## S3 method for class 'deconv'
cooks.distance(model, ...)

Arguments

Details

Residuals are first adjusted for gene weights (if used). rstandard and rstudent give standardized and Studentized residuals respectively. Standardised residuals are calculated based on the hat matrix:

H = X (X^T X)^{-1} X^T

Leverage h_{ii} = diag(H) is used to standardise the residuals:

t_i = \cfrac{\hat{\varepsilon_i}}{\hat{\sigma} \sqrt{1 - h_{ii}}}

Studentized residuals are calculated based on excluding the i th case. Note this corresponds to refitting the regression, but without recomputing the non-negative compensation matrix. Cook's distance is calculated as:

D_i = \cfrac{e_i^2}{ps^2} \left[\cfrac{h_{ii}}{(1 - h_{ii})^2} \right]

where p is the number of predictors (cell subclasses) and s^2 is the mean squared error. In this model the intercept is not included.

Value

Matrix of adjusted residuals or Cook's distance.

See Also

stats::influence.measures()

Single-cell apply a function to a matrix split by a factor

Description

Workhorse function designed to handle large scRNA-Seq gene expression matrices such as embedded Seurat matrices, and apply a function to columns of the matrix split as a ragged array by an index factor, similar to tapply(), by() or aggregate(). Note that here the index is applied to columns as these represent cells in the single-cell format, rather than rows as in aggregate(). Very large matrices are handled by slicing rows into blocks to avoid excess memory requirements.

Usage

scapply(
  x,
  INDEX,
  FUN,
  combine = NULL,
  combine2 = "c",
  progress = TRUE,
  sliceMem = 16,
  cores = 1L,
  ...
)

Arguments

Details

The limit on sliceMem is that the number of elements manipulated in each block must be kept below the long vector limit of 2^31 (around 2e9). Increasing cores requires substantial amounts of spare RAM. combine works in a similar way to .combine in foreach(); it works across the levels in INDEX. combine2 is nested and works across slices of genes (an inner loop), so it is only invoked if slicing occurs which is when a matrix has a larger memory footprint than sliceMem.

Value

By default returns a list, unless combine is invoked in which case the returned data type will depend on the functions specified by FUN and combine.

Author(s)

Myles Lewis

See Also

scmean() which applies a fixed function logmean() in a similar manner, and slapply() which applies a function to a big matrix with slicing but without splitting by an index factor.

Examples

# equivalent
m <- matrix(sample(0:100, 1000, replace = TRUE), nrow = 10)
cell_index <- sample(letters[1:5], 100, replace = TRUE)
o <- scmean(m, cell_index)
o2 <- scapply(m, cell_index, function(x) rowMeans(log2(x +1)),
              combine = "cbind")
identical(o, o2)

Single-cell mean log gene expression across cell types

Description

Workhorse function which takes as input a scRNA-Seq gene expression matrix such as embedded in a Seurat object, calculates log2(counts +1) and averages gene expression over a vector specifying cell subclasses or cell types. Very large matrices are handled by slicing rows into blocks to avoid excess memory requirements.

Usage

scmean(
  x,
  celltype,
  FUN = "logmean",
  postFUN = NULL,
  verbose = TRUE,
  sliceMem = 16,
  cores = 1L,
  load_balance = FALSE,
  use_future = FALSE
)

Arguments

Details

Mean functions which can be applied by setting FUN include logmean (the default) which applies row means to log2(counts+1), or trimmean which calculates the trimmed mean of the counts after top/bottom 5% of values have been excluded. Alternatively FUN = rowMeans calculates the arithmetic mean of counts.

If FUN = trimmean or rowMeans, postFUN needs to be set to log2s which is a simple function which applies log2(x+1).

sliceMem can be set lower on machines with less RAM, but this will slow the analysis down. cores increases the theoretical amount of memory required to around cores * sliceMem in GB. For example on a 64 GB machine, we find a significant speed increase with cores = 3L. Above this level, there is a risk that memory swap will slow down processing.

Value

a matrix of mean log2 gene expression across cell types with genes in rows and cell types in columns.

Author(s)

Myles Lewis

See Also

scapply() which is a more general version which can apply any function to the matrix. logmean, trimmean are options for controlling the type of mean applied.

Gene signature heatmap

Description

Produces a heatmap of genes signatures for each cell subclass using ComplexHeatmap.

Usage

signature_heatmap(
  x,
  type = c("subclass", "group", "groupsplit"),
  top = Inf,
  use_filter = NULL,
  arith_mean = FALSE,
  rank = c("max", "angle"),
  scale = c("none", "max", "sphere"),
  col = rev(hcl.colors(10, "Greens3")),
  text = TRUE,
  fontsize = 6.5,
  outlines = FALSE,
  outline_col = "black",
  subset = NULL,
  add_genes = NULL,
  ...
)

Arguments

Value

A 'Heatmap' class object.

Simulate pseudo-bulk RNA-Seq

Description

Simulates pseudo-bulk RNA-Seq dataset using two modes. The first mode uses a 'cellMarkers' class object and a matrix of counts for the numbers of cells of each cell subclass. This method converts the log2 gene means back for each cell subclass back to count scale and then calculates pseudo-bulk count values based on the cell amounts specified in samples. In the 2nd mode, a single-cell RNA-Seq dataset is required, such as a matrix used as input to cellMarkers(). Cells from the relevant subclass are sampled from the single-cell matrix in the appropriate amounts based on samples, except that sampling is scaled up by the factor times.

Usage

simulate_bulk(
  object,
  samples,
  subclass,
  times = 1,
  method = c("dirichlet", "unif"),
  alpha = 1
)

Arguments

Details

The first method can give perfect deconvolution if the following settings are used with deconvolute(): count_space = TRUE, convert_bulk = FALSE, use_filter = FALSE and comp_amount = 1.

Value

An integer count matrix with genes in rows and cell subclasses in columns. This can be used as test with the deconvolute() function.

See Also

generate_samples() deconvolute() add_noise()

Apply a function to a big matrix by slicing

Description

Workhorse function ('slice apply') designed to handle large scRNA-Seq gene expression matrices such as embedded Seurat matrices, and apply a function to the whole matrix. Very large matrices are handled by slicing rows into blocks to avoid excess memory requirements.

Usage

slapply(x, FUN, combine = "c", progress = TRUE, sliceMem = 16, cores = 1L, ...)

Arguments

Details

The limit on sliceMem is that the number of elements manipulated in each block must be kept below the long vector limit of 2^31 (around 2e9). Increasing cores requires substantial amounts of spare RAM. combine works in a similar way to .combine in foreach() across slices of genes; it is only invoked if slicing occurs.

Value

The returned data type will depend on the functions specified by FUN and combine.

Author(s)

Myles Lewis

See Also

scapply()

Specificity plot

Description

Scatter plot showing specificity of genes as markers for a particular cell subclass. Optimal gene markers for that cell subclass are those genes which are closest to or lie on the y axis, while also being of highest mean expression.

Usage

specificity_plot(
  mk,
  subclass = NULL,
  group = NULL,
  type = 1,
  use_filter = FALSE,
  nrank = 8,
  nsubclass = NULL,
  expfilter = NULL,
  scheme = NULL,
  add_labels = NULL,
  label_pos = "right",
  axis_extend = 0.4,
  nudge_x = NULL,
  nudge_y = NULL,
  ...
)

specificity_plotly(
  mk,
  subclass = NULL,
  group = NULL,
  type = 1,
  use_filter = FALSE,
  nrank = 8,
  nsubclass = NULL,
  expfilter = NULL,
  scheme = NULL,
  ...
)

Arguments

Details

For type = 1, coordinates are drawn as x = angle of vector in degrees, y = mean gene expression of each gene in the subclass of interest. This version is easier to use to identify additional gene markers. The plotly version allows users to hover over points and identify which gene they belong to.

If type = 2, the coordinates are drawn as x = vector length * sin(angle) and y = vector length * cos(angle), where vector length is the Euclidean length of that gene in space where each cell subclass is a dimension. Angle is the angle between the projected vector in space against perfection for that cell subclass, i.e. the vector lying perfectly along the subclass dimension with no deviation along other subclass dimensions, i.e. a gene which is expressed solely in that subclass and has 0 expression in all other subclasses. y is equal to the mean expression of each gene in the subclass of interest. x represents the Euclidean distance of mean expression in all other subclasses, i.e. overall non-specific gene expression in other subclasses. Thus, the plot represents a rotation of all genes as vectors around the axis of the subclass of interest onto the same plane so that the angle with the subclass of interest is visualised between genes.

Colour is used to overlay the ranking of each gene across the subclasses, showing for each gene where the subclass of interest is ranked compared to the other subclasses. Best markers have the subclass of interest ranked 1st.

Value

ggplot2 or plotly scatter plot object.

Spillover heatmap

Description

Produces a heatmap from a 'cellMarkers' or 'deconv' class object showing estimated amount of spillover between cell subclasses. The amount that each cell subclass's overall vector spillovers (projects) into other cell subclasses' vectors is shown in each row. Thus the column gives an estimate of how much the most influential (specific) genes for a cell subclass are expressed in other cells.

Usage

spillover_heatmap(
  x,
  text = NULL,
  cutoff = 0.5,
  fontsize = 8,
  subset = NULL,
  ...
)

Arguments

Value

No return value. Draws a heatmap using ComplexHeatmap.

Stacked bar plot

Description

Produces stacked bar plots using base graphics or ggplot2 showing amounts of cell subclasses in deconvoluted bulk samples.

Usage

stack_plot(
  x,
  percent = FALSE,
  order_col = 1,
  scheme = NULL,
  order_cells = c("none", "increase", "decrease"),
  seriate = NULL,
  cex.names = 0.7,
  show_xticks = TRUE,
  ...
)

stack_ggplot(
  x,
  percent = FALSE,
  order_col = 1,
  scheme = NULL,
  order_cells = c("none", "increase", "decrease"),
  seriate = NULL,
  legend_ncol = NULL,
  legend_position = "bottom",
  show_xticks = FALSE
)

Arguments

Value

The base graphics function has no return value. It plots a stacked barchart using base graphics. The ggplot2 version returns a ggplot2 object.

Summarising deconvolution tuning

Description

summary method for class 'tune_deconv'.

Usage

## S3 method for class 'tune_deconv'
summary(
  object,
  metric = attr(object, "metric"),
  method = attr(object, "method"),
  ...
)

Arguments

Value

If method = "top" prints the row representing the best tuning of parameters (maximum mean R squared, averaged across subclasses). For method = "overall", the average effect of varying each parameter is calculated by mean R-squared across the rest of the grid and the best value for each parameter is printed. Invisibly returns a dataframe of mean metric values (Pearson r^2, R^2, RMSE) averaged over subclasses.

Tune deconvolution parameters

Description

Performs an exhaustive grid search over a tuning grid of cell marker and deconvolution parameters for either updateMarkers() (e.g. expfilter or nsubclass) or deconvolute() (e.g. comp_amount).

Usage

tune_deconv(
  mk,
  test,
  samples,
  grid,
  output = "output",
  metric = "RMSE",
  method = "top",
  verbose = TRUE,
  cores = 1,
  ...
)

Arguments

Details

Tuning plots on the resulting object can be visualised using plot_tune(). If best_tune is set to "overall", this corresponds to setting subclass = NULL in plot_tune().

Once the results output has been generated, arguments such as metric or method can be changed to see different best tunes using summary() (see summary.tune_deconv()).

test and samples matrices can be generated by simulate_bulk() and generate_samples() based on the original scRNA-Seq count dataset.

Value

Dataframe with class 'tune_deconv' whose columns include: the parameters being tuned via grid, cell subclass and R squared.

See Also

plot_tune() summary.tune_deconv()

Update cellMarkers object

Description

Updates a 'cellMarkers' gene signature object with new settings without having to rerun calculation of gene means, which can be slow.

Usage

updateMarkers(
  object = NULL,
  genemeans = NULL,
  groupmeans = NULL,
  add_gene = NULL,
  add_groupgene = NULL,
  remove_gene = NULL,
  remove_groupgene = NULL,
  remove_subclass = NULL,
  remove_group = NULL,
  bulkdata = NULL,
  nsubclass = object$opt$nsubclass,
  ngroup = object$opt$ngroup,
  expfilter = object$opt$expfilter,
  noisefilter = object$opt$noisefilter,
  noisefraction = object$opt$noisefraction,
  verbose = TRUE
)

Arguments

Value

A list object of S3 class 'cellMarkers'. See cellMarkers() for details. If gene2symbol() has been called, an extra list element symbol will be present. The list element update stores the call to updateMarkers().

Author(s)

Myles Lewis

See Also

cellMarkers() gene2symbol()

Cell subclass violin plot

Description

Produces violin plots using ggplot2 showing amounts of cell subclasses in deconvoluted bulk samples.

Usage

violin_plot(x, percent = FALSE, order_cols = c("none", "increase", "decrease"))

Arguments

Value

A ggplot2 plotting object.