runspatialdecon: Run spatialdecon

In Nanostring-Biostats/SpatialDecon: Deconvolution of mixed cells from spatial and/or bulk gene expression data

Run spatialdecon

Description

Runs spatialdecon from an S4 object

A wrapper for applying spatialdecon to a NanostringGeomxSet object.

A wrapper for applying spatialdecon to the Spatial data element in a Seurat object. Unlike spatialdecon, which expects a normalized data matrix, this function operates on raw counts. Scaling for total cells

Usage

runspatialdecon(object, ...)

## S4 method for signature 'NanoStringGeoMxSet'
runspatialdecon(
  object,
  X = NULL,
  norm_elt = NULL,
  raw_elt = NULL,
  wts = NULL,
  resid_thresh = 3,
  lower_thresh = 0.5,
  align_genes = TRUE,
  is_pure_tumor = NULL,
  n_tumor_clusters = 10,
  cell_counts = NULL,
  cellmerges = NULL,
  maxit = 1000
)

## S4 method for signature 'Seurat'
runspatialdecon(
  object,
  X = NULL,
  bg = 0.1,
  wts = NULL,
  resid_thresh = 3,
  lower_thresh = 0.5,
  align_genes = TRUE,
  is_pure_tumor = NULL,
  n_tumor_clusters = 10,
  cell_counts = NULL,
  cellmerges = NULL,
  maxit = 1000
)

Arguments

object	An S4 object such as a Seurat object that includse a "Spatial" element in the "assays" slot or a GeoMxSet object
...	Arguments passed to spatialdecon
X	Cell profile matrix. If NULL, the safeTME matrix is used.
norm_elt	normalized data element in assayData in NanostringGeomxSet object
raw_elt	raw data element in assayData in NanostringGeomxSet object
wts	Optional, a matrix of weights.
resid_thresh	A scalar, sets a threshold on how extreme individual data points' values can be (in log2 units) before getting flagged as outliers and set to NA.
lower_thresh	A scalar. Before log2-scale residuals are calculated, both observed and fitted values get thresholded up to this value. Prevents log2-scale residuals from becoming extreme in points near zero.
align_genes	Logical. If TRUE, then Y, X, bg, and wts are row-aligned by shared genes.
is_pure_tumor	A logical vector denoting whether each AOI consists of pure tumor. If specified, then the algorithm will derive a tumor expression profile and merge it with the immune profiles matrix.
n_tumor_clusters	Number of tumor-specific columns to merge into the cell profile matrix. Has an impact only when is_pure_tumor argument is used to indicate pure tumor AOIs. Takes this many clusters from the pure-tumor AOI data and gets the average expression profile in each cluster. Default 10.
cell_counts	Number of cells estimated to be within each sample. If provided alongside norm_factors, then the algorithm will additionally output cell abundance esimtates on the scale of cell counts.
cellmerges	A list object holding the mapping from beta's cell names to combined cell names. If left NULL, then defaults to a mapping of granular immune cell definitions to broader categories.
maxit	Maximum number of iterations. Default 1000.
bg	Expected background counts. Either a scalar applied equally to all points in the count matrix, or a matrix with the same dimensions as the count matrix in GetAssayData(object, assay = "Spatial"). Recommended to use a small non-zero value, default of 0.1.

Value

decon results in list or in GeoMxSet object

For GeoMxSet object, if not given cellmerges and cell_counts, a valid GeoMx S4 object including the following items

• In pData

  • beta: matrix of cell abundance estimates, cells in rows and observations in columns

  • p: matrix of p-values for H0: beta == 0

  • t: matrix of t-statistics for H0: beta == 0

  • se: matrix of standard errors of beta values

  • prop_of_all: rescaling of beta to sum to 1 in each observation

  • prop_of_nontumor: rescaling of beta to sum to 1 in each observation, excluding tumor abundance estimates

  • sigmas: covariance matrices of each observation's beta estimates

• In assayData

  • yhat: a matrix of fitted values

  • resids: a matrix of residuals from the model fit. (log2(pmax(y, lower_thresh)) - log2(pmax(xb, lower_thresh))).

• In experimentData

  • SpatialDeconMatrix: the cell profile matrix used in the decon fit.

if given cellmerges, the valid GeoMx S4 object will additionally include the following items

• In pData

  • beta.granular: cell abundances prior to combining closely-related cell types

  • sigma.granular: sigmas prior to combining closely-related cell types

if given cell_counts, the valid GeoMx S4 object will additionally include the following items

• In pData

  • cell.counts: beta rescaled to estimate cell numbers, based on prop_of_all and nuclei count

if given both cellmerges and cell_counts, the valid GeoMx S4 object will additionally include the following items

• In pData

  • cell.counts.granular: cell.counts prior to combining closely-related cell types

For Seurat Object, if not given cellmerges and cell_counts, a list including the following items:

• beta: matrix of cell abundance estimates, cells in rows and observations in columns

• p: matrix of p-values for H0: beta == 0

• t: matrix of t-statistics for H0: beta == 0

• se: matrix of standard errors of beta values

• prop_of_all: rescaling of beta to sum to 1 in each observation

• prop_of_nontumor: rescaling of beta to sum to 1 in each observation, excluding tumor abundance estimates

• yhat: a matrix of fitted values

• resids: a matrix of residuals from the model fit. (log2(pmax(y, lower_thresh)) - log2(pmax(xb, lower_thresh))).

• X: the cell profile matrix used in the decon fit.

• sigmas: covariance matrices of each observation's beta estimates

if given cellmerges, the list will additionally include the following items

• beta.granular: cell abundances prior to combining closely-related cell types

• sigma.granular: sigmas prior to combining closely-related cell types

if given cell_counts, the list will additionally include the following items

• cell.counts: beta rescaled to estimate cell numbers, based on prop_of_all and nuclei count

if given both cellmerges and cell_counts, the list will additionally include the following items

• cell.counts.granular: cell.counts prior to combining closely-related cell types

Examples

## GeoMxSet Object ##
library(GeomxTools)
datadir <- system.file("extdata", "DSP_NGS_Example_Data", package = "GeomxTools")
demoData <- readRDS(file.path(datadir, "/demoData.rds"))

demoData <- shiftCountsOne(demoData)
target_demoData <- aggregateCounts(demoData)

target_demoData <- normalize(target_demoData, "quant")

demoData <- runspatialdecon(object = target_demoData, 
                            norm_elt = "exprs_norm",
                            raw_elt = "exprs")

## Seurat Object ##
# get dataset
con <- gzcon(url("https://github.com/almaan/her2st/raw/master/data/ST-cnts/G1.tsv.gz"))
txt <- readLines(con)
temp <- read.table(textConnection(txt), sep = "\t", header = TRUE, row.names = 1)
# parse data
raw = t(as.matrix(temp))
norm = sweep(raw, 2, colSums(raw), "/") * mean(colSums(raw))
x = as.numeric(substr(rownames(temp), 1, unlist(gregexpr("x", rownames(temp))) - 1))
y = -as.numeric(substr(rownames(temp), 
                 unlist(gregexpr("x", rownames(temp))) + 1, nchar(rownames(temp))))
# put into a seurat object:
andersson_g1 = SeuratObject::CreateSeuratObject(counts = raw, assay="Spatial")
[e-mail]$x = x
[e-mail]$y = y

res <- runspatialdecon(andersson_g1)
str(res)

Nanostring-Biostats/SpatialDecon documentation built on Jan. 26, 2024, 8:20 p.m.