Package 'cytosignal'

Title: What the Package Does (One Line, Title Case)
Description: What the package does (one paragraph).
Authors: Jialin Liu [aut, cre], Yichen Wang [aut]
Maintainer: Jialin Liu <[email protected]>
License: GPL (>= 3)
Version: 0.2.0
Built: 2024-09-14 05:43:46 UTC
Source: https://github.com/welch-lab/cytosignal

Help Index

#' Compute the LR velo for specific ligand-receptor imputation obj pairs #' #' @param object A Cytosignal object #' @param lig.slot The ligand slot to use #' @param recep.slot The receptor slot to use #' @param intr.db.name The intr database name to use #' @param nn.use The neighbor index as niche #' #' @return A Cytosignal object #' @export #' inferVeloLR <- function( object, ... ) UseMethod(generic = 'inferVeloLR', object = object)

Description

#' Compute the LR velo for specific ligand-receptor imputation obj pairs #' #' @param object A Cytosignal object #' @param lig.slot The ligand slot to use #' @param recep.slot The receptor slot to use #' @param intr.db.name The intr database name to use #' @param nn.use The neighbor index as niche #' #' @return A Cytosignal object #' @export #' inferVeloLR <- function( object, ... ) UseMethod(generic = 'inferVeloLR', object = object)

Usage

.inferVeloLR.matrix_like(
  dge.lig,
  dge.recep,
  dge.lig.velo,
  dge.recep.velo,
  lig.fac,
  recep.fac
)

Add interaction database to CytoSignal object

Description

Add interaction database to CytoSignal object

Usage

addIntrDB(
  object,
  gene_to_uniprot,
  intr.db.diff_dep,
  intr.db.cont_dep,
  inter.index
)

Arguments

object

CytoSignal object
gene_to_uniprot

df, gene to uniprot symbols mapping
intr.db.diff_dep

intr.db for diffusable ligands + non-diffusable receptors
intr.db.cont_dep

intr.db for contact dependent ligands + receptors
inter.index

df, collection of intraction information

Value

a CytoSignal object

Add velocity data to CytoSignal object

Description

Add velocity data to CytoSignal object

Usage

addVelo(object, velo.s, velo.u)

Arguments

object

CytoSignal object
velo.s

matrix of spliced velo
velo.u

matrix of unspliced velo

Subset gene expression matrix according to availability in the UNIPROT database.

Description

Subset genes to those available in g_to_u, the gene name to UNIPROT ID databse provided with addIntrDB

Usage

changeUniprot(object, ...)

## S3 method for class 'matrix_like'
changeUniprot(object, gene_to_uniprot, verbose = TRUE, ...)

## S3 method for class 'CytoSignal'
changeUniprot(object, verbose = TRUE, ...)

Arguments

object

A CytoSignal object or a matrix.
...

Parameters passed to S3 methods.
gene_to_uniprot

Please use g_to_u.
verbose

Whether to show information of the progress.

Value

CytoSignal object with subset matrix updated in counts slot, or the subset matrix.

Create a CytoSignal object

Description

Create a CytoSignal object

Usage

createCytoSignal(
  raw.data,
  cells.loc,
  clusters = NULL,
  name = NULL,
  parameters = NULL,
  log = NULL
)

Arguments

raw.data

raw data matrix
cells.loc

cells location matrix
clusters

cluster information
name

name of the dataset
parameters

parameters used
log

log of the processing

Value

a CytoSignal object

The CytoSignal Class

Description

The CytoSignal object is created from one spatial transcriptomic dataset. A CytoSignal object can be created with createCytoSignal.

Slots

counts

Raw gene expression count matrix, sparse dgCMatrix, gene x cell.

cell.loc

Matrix of cell locations, cell by axis.

clusters

cluster assignments, named factor.

imputation

List of imputation objects.

lrscore

List of lrScore objects.

velo

List of velo objects.

intr.valid

List of interaction database. Added with addIntrDB.

cell.data

data.frame of cell data.

parameters

List of parameters for gaussian imputation

log

Log of each main steps

version

Version of package used to create object. Should not be modified by users.

Interaction database derived from CellphoneDB V2

Description

The referencing database that contains the list of curated interactions, the meta-information of each interaction, the ligand and receptor of each interaction, the protein components of both part of each interaction. The protein components are presented with UNIPROT ID, so we also provide g_to_u that maps UNIPROT ID to gene names.

Usage

db.diff

db.cont

inter.index

g_to_u

Format

db.diff and db.cont are lists of factor objects. db.diff is for diffusion dependent interactions and db.cont is for contact dependent interactions. Both lists have the following three entries of the same format:

  • $ligands - A factor object where the values are IDs of interactions and names are UNIPROT IDs of the ligand protein components that match to the interaction.

  • $receptors - A factor object of the same format as $ligands but for the receptor part of the interactions.

  • $combined - The concatenation of the two factors above.

inter.index is a data.frame object with the meta-information of each interaction.

g_to_u is a data.frame object that map gene names to UNIPROT ID of their protein product.

An object of class list of length 3.

An object of class data.frame with 1396 rows and 11 columns.

An object of class data.frame with 977 rows and 3 columns.

Source

CellphoneDB V2

Identify nearest neighbors for each location using different strategies for different types of interactions

Description

This is a wrapper function of findNNGauEB, findNNDT and findNNRaw. We use a Gaussian Epsilon ball to identify the nearest neighbors that can contribute to the diffusible ligands. And then uses a Delaunay triangulation to identify the nearest neighbors that can contribute to the contact-dependent ligands. The identified nearest neighbors will then be used for imputing the LL or RR value for each location.

Usage

findNN(
  object,
  eps = NULL,
  sigma = NULL,
  diff.weight = "auto",
  dt.weight = 2,
  max.r = NULL
)

Arguments

object

A CytoSignal object.
eps

Gaussian Epsilon Ball method parameter. A numeric scalar. The radius of the Gaussian Epsilon ball. Default NULL uses parameter inferred with inferEpsParams.
sigma

Gaussian Epsilon Ball method parameter. The σ\sigma of the Gaussian kernel. Default NULL uses parameter inferred with inferEpsParams.
max.r

Delaunay Triangulation method parameter. A numeric scalar for the maximum radius of the edges. Default NULL uses parameter inferred with inferEpsParams.
self.weight

Gaussian Epsilon Ball method parameter. Weight of the index cell. Use a number between 0-1 or a string "auto" or "sum_1". Default "auto".
weight

Delaunay Triangulation method parameter. A numeric scalar for the sum of the weights of the edges of the Delaunay Triangulation. Default 2.

Value

A CytoSignal object updated. object@imputation slot will be updated with three new entries: object@imputation$GauEps, object@imputation$DT and object@imputation$Raw.

Examples

## Not run: 
object <- findNN(object)

## End(Not run)

Find the direct connected neighbor of each cell, using Delaunay triangulation

Description

Find the direct connected neighbor of each cell, using Delaunay triangulation

Usage

findNNDT(object, ...)

## S3 method for class 'matrix'
findNNDT(object, weight.sum = 2, max.r = NULL, ...)

## S3 method for class 'CytoSignal'
findNNDT(object, weight = 2, max.r = NULL, ...)

Arguments

object

A CytoSignal object or a matrix of cell location.
...

Arguments passed to other S3 methods
weight.sum

The sum of the weights
max.r

The maximum radius of the edges, by default is the r.diffuse.scale,

Value

For CytoSignal object, the original object with result updated. For matrix object, a list of neighbors.

Find the neighbors of each cell in the Epsilon Ball

Description

Find the neighbors of each cell in the Epsilon Ball

Usage

findNNGauEB(object, eps = NULL, sigma = NULL, self.weight = "auto", ...)

## S3 method for class 'matrix'
findNNGauEB(object, eps, sigma = 0.15, self.weight = "auto", ...)

## S3 method for class 'CytoSignal'
findNNGauEB(
  object,
  eps = NULL,
  sigma = NULL,
  self.weight = "auto",
  tag = NULL,
  ...
)

Arguments

object

A CytoSignal object or a matrix object for the cell spatial location.
eps

The radius of the epsilon ball. For CytoSignal object, by default use parameter inferred with inferEpsParams.
sigma

The sigma of the Gaussian kernel. Default 0.15 for matrix method. For CytoSignal object, by default use parameter inferred with inferEpsParams.
self.weight

weight of the index cell. Use a number between 0-1, or choose from "auto" or "sum_1".
...

Arguments passed to other S3 methods.
tag

Name prefix of the analysis.

Value

For CytoSignal object, the orignal object with result updated. For matrix object, a list of neighbors and their distances.

Create a ImpData object using raw data without imputation

Description

Create a ImpData object using raw data without imputation

Usage

findNNRaw(object)

Arguments

object

A Cytosignal object

Value

A Cytosignal object

Format a CytosignalIntrDEG object to string

Description

Format a CytosignalIntrDEG object to string

Usage

## S3 method for class 'CytosignalIntrDEG'
format(x, ...)

Arguments

x

A CytosignalIntrDEG object.
...

Passed to other methods

Value

A string representation of the object

Compute the LR score for specific ligand-receptor imputation obj pairs

Description

Compute the LR score for specific ligand-receptor imputation obj pairs

Usage

graphNicheLR(object, ...)

Arguments

object

A Cytosignal object
lig.slot

The ligand slot to use
recep.slot

The receptor slot to use
intr.db.name

The intr database name to use
nn.use

The neighbor index as niche

Value

A Cytosignal object

Sub function for graphNicheLR, input a CytoSignal object

Description

Sub function for graphNicheLR, input a CytoSignal object

Sub function for graphNicheLR, input a CytoSignal object

Usage

## S3 method for class 'CytoSignal'
graphNicheLR(object, lig.slot, recep.slot, intr.db.name, nn.use = NULL)

## S3 method for class 'CytoSignal'
graphNicheLR(object, lig.slot, recep.slot, intr.db.name, nn.use = NULL)

Arguments

object

A Cytosignal object
lig.slot

The ligand slot to use
recep.slot

The receptor slot to use
intr.db.name

The intr database name to use
nn.use

slot that the neighbor index should be taken from, by default is the same as the recep.slot. For example, if score.obj = GauEps-DT, then nn.use = "DT". nn.use could also be a user-defind factor.

Value

A Cytosignal object

A Cytosignal object

Sub function for graphNicheLR, input a sparse matrix

Description

Sub function for graphNicheLR, input a sparse matrix

Usage

## S3 method for class 'dgCMatrix'
graphNicheLR(dge.lig, dge.recep, nb.id.fac, lig.fac, recep.fac)

Arguments

dge.lig

A sparse matrix for ligand
dge.recep

A sparse matrix for receptor
nb.id.fac

A factor of neighbor indices
lig.fac

A factor of ligand indices
recep.fac

A factor of receptor indices

Value

A sparse matrix

Show significant genes across top GO term hits with coefficients from regression analysis of an interaction

Description

Create a heatmap for an interaction on GO terms by significant genes, colored by the coefficients of the genes in the regression model returned by inferIntrDEG. The GO term enrichment can be done with any tools available. The input data.frame GO must contain fields for 1. term description, character, pointed to by description.col, 2. p-value, numeric, pointed to by pval.col, 3. gene hit string, character, pointed to by gene.col. The gene hit string for each term must be form in a way that function gene.split.fun can split it into a character vector of gene names. For example, if a gene hit string is "gene1, gene2, gene3", then gene.split.fun should be function(x) unlist(strsplit(x, ", ")), so that a split result c("gene1", "gene2", "gene3") can be correctly obtained.

Usage

heatmap_GO(
  intrDEG,
  GO,
  intr,
  description.col = "description",
  pval.col = "pval",
  gene.col = "genes",
  gene.split.fun = function(x) unlist(strsplit(x, ",")),
  term.topN = 20,
  gene.topN = 20,
  binary_sign = FALSE,
  text.size = 10
)

Arguments

intrDEG

A CytosignalIntrDEG object, output from inferIntrDEG.
GO

A data.frame object for GO enrichment analysis result.
intr

A single interaction ID (starts with "CPI") or its numeric index within the range of intrDEG.
description.col, pval.col, gene.col

The column names of the data.frame GO that contains the term description, p-value, and gene hit string. Default "description", "pval", "genes".
term.topN

Use this number of top GO terms, ranked by p-values. Default 20.
gene.topN

Use this number of top genes, ranked by absolute value of coefficients. Default 20.
binary_sign

Whether to convert coefficient value to binary sign value. Default FALSE.
color_num

Number of colors in the heatmap. Can only use 2 or 3, Default 2 use white-red color palette. 3 use scaled blue-white-red color palette.

GGPLOT2 FUNCTIONALITY FOR MAPPING TO HEXAGON SIZE AND COLOUR AESTHETICS by Robin Edwards, 2013 (geotheory.co.uk, @geotheory) This has been adapted from the ggplot bin_hex.R script that underpins geom_hex, etc (see https://github.com/hadley/densityvis/blob/master/R/bin-hex.r).

Description

These functions implement aesthetic mapping to hexagon size (area), in addition to the existing colour-mapping functionality. The key change is the addition of a new fourth variable (var4) to hex_bin(), which complements the inbuilt hexagon binning functionality. The 'frequency.to.area' argument enables the default mappings of binned data to colour and var4 to size to be interchanged. The hmin/hmax arguments 0,1 set area mapping constraints (hmax can exceed 1). xlim/xlat enable hexagon tesselation to be constrained independently of data range. There may be some bugs in the implementation. A legend for hexagon size has not been implemented. Bin data into hexagons (2d).

Usage

hex_bin(
  x,
  y,
  weight = NULL,
  var4 = NULL,
  width = NULL,
  height = NULL,
  xbins = 20,
  ybins = 20,
  var4.to.color = FALSE,
  na.rm = FALSE,
  hmin = 0,
  hmax = 1,
  xlim = NULL,
  ylim = NULL,
  ...
)

Arguments

x

a numeric vector of x positions
y

a numeric vector of y positions
weight

NULL or a numeric vector providing weights for each observation, replace counts, mapped to color
var4

NULL or a numeric vector providing weights for each observation, averaged within each bin, mapped to hex bin size
width

width of each hexagon, if NULL computed from ybins
height

height of each hexagon, if NULL computed from ybins
xbins

number of horizontal bins, if width unspecified
ybins

number of vertical bins, if height unspecified
na.rm

If TRUE missing values will be silently removed, otherwise they will be removed with a warning.

Value

A data frame with columns x, y and freq, and attributes width and height.

See Also

hex_pos for algorithm that finds hexagon center closest to each point and hex_coord that generates coordinates of each hexagon.

Examples

plot(hex_bin(runif(1e4), runif(1e4)))
plot(hex_bin(rnorm(1e4), rnorm(1e4)))

data(baseball, package = "plyr")
bin <- hex_bin(baseball$g, baseball$ab)

The ImpData Class

Description

The ImpData object is created from one ST dataset. User could choose a preferred imputation method and the class stores the imputed data, the imputed normalized data, the intr database, the intr database.

Details

The key slots used in the ImpData object are described below.

Slots

method

imputation method

imp.data

imputed data

imp.norm.data

imputed normalized data

intr.data

imputed normalized data subsetted by intr database

intr.data.null

permuted imputed normalized data subsetted by intr database

nn.id

nearest neighbor id

nn.dist

nearest neighbor distance

log

Log of each main steps

Impute the LL or RR value from the nearest neighbors of each location

Description

After running findNN, we can impute the LL or RR value from the nearest neighbors of each location basing on the types of nearest neighbors.

Usage

imputeLR(object, weights = c("none", "mean", "counts", "dist"))

Arguments

object

A CytoSignal object, with findNN already run.
weights

The method to transform distances to weights. Choose from "none", "mean", "counts" and "dist". Default "none" since weights are already calculated by Gaussian kernel and DT mean.

Value

A CytoSignal object updated. Entries in object@imputation slot will be updated with the imputation values.

Examples

## Not run: 
object <- findNN(object)
object <- imputeLR(object)

## End(Not run)

Impute the data

Description

Impute the data

Usage

imputeNiche(object, weights = c("mean", "counts", "dist", "none"), ...)

## S3 method for class 'dgCMatrix'
imputeNiche(
  object,
  nb.id.fac,
  nb.dist.fac,
  weights = c("mean", "counts", "dist", "none"),
  ...
)

## S3 method for class 'CytoSignal'
imputeNiche(
  object,
  nn.type = NULL,
  weights = c("mean", "counts", "dist", "none"),
  ...
)

Arguments

object

A Cytosignal object or a dgCMatrix object of raw gene expression matrix
weights

The weight of the Delaunay triangulation. Choose from "mean", "counts", "dist" or "none".
...

Arguments passed to other S3 methods
nb.id.fac

A factor of neighbors
nb.dist.fac

A factor of weights
nn.type

The type of neighbors. The tag used when finding nearest neighbors in the upstream step. If not customized, use "GauEps" for NN found with findNNGauEB, or use "DT" for NN found with findNNDT. Default use the most lastly computed NN.

Value

For CytoSignal object, the original object with imputation result updated. For dgCMatrix object, a sparse N x N graph presented as another dgCMatrix object.

Impute the velocity mtx using the specified method

Description

Impute the velocity mtx using the specified method

Usage

imputeNicheVelo(object, ...)

Arguments

object

A Cytosignal object
...

Other parameters
method

The method to use for imputation

Value

A Cytosignal object

Sub function for imputeNicheVelo, input a Cytosignal object

Description

Sub function for imputeNicheVelo, input a Cytosignal object

Usage

## S3 method for class 'CytoSignal'
imputeNicheVelo(object, nn.type = NULL)

Arguments

object

A Cytosignal object
nn.type

The type of neighbors

Value

A Cytosignal object

Impute time derivative of LL or RR from the nearest neighbors of each location

Description

After running findNN, we can impute the temporal LL or RR change from the nearest neighbors of each location basing on the types of nearest neighbors.

Usage

imputeVeloLR(object)

Arguments

object

A CytoSignal object, with findNN already run and velocity information added with addVelo.

Value

A CytoSignal object updated. Entries in object@imputation slot will be updated with the imputation values.

Examples

## Not run: 
object <- findNN(object)
object <- imputeVeloLR(object)

## End(Not run)

Infer the correspondence between LR-scores and Significance

Description

Min-max the LRscores, substract the significant ones, sum and average.

Usage

inferCorrScore(object, correctBy = c("cell", "intr"), slot.use = NULL)

Arguments

object

A Cytosignal object
correctBy

Spatial FDR correction method. Choose from "cell" or "intr".
slot.use

Which LR score to use. Use the name specified with tag when running inferLRScore.

Value

A Cytosignal object

Infer the parameters of the Gaussian kernel

Description

Infer the parameters of the Gaussian kernel

Usage

inferEpsParams(object, scale.factor = NULL, r.eps.real = 200, thresh = 0.001)

Arguments

object

A Cytosignal object
r.eps.real

The radius of the epsilon ball in tech resolution in um, default 200 um
thresh

The total signal out of the epsilon ball
scale.facor

1 spatial coord unit equals to how many µm

Value

A Cytosignal object

Infer significant genes for each interaction

Description

This function performs wilcoxon one-sided test for each cell type, between cells enriched with an interaction against other cells. The top significant genes are selected for each interaction. The expression profile of the selected genes, together with the cell type variable, are then feeded into a regression model for further refinement. This analysis infers the significant genes for each interaction, and refines the LRScore using the selected genes.

Usage

inferIntrDEG(
  object,
  intr = NULL,
  slot.use = NULL,
  signif.use = NULL,
  num.per.choose = 50,
  alpha.test = seq(0.5, 1, 0.1),
  seed = 1,
  minCell = 50,
  verbose = TRUE
)

Arguments

object

A Cytosignal object with inferIntrScore already run.
intr

Specify interactions to be used. A vector of either the unique ID of interactions or the numeric rank indices. Available IDs can be shown with showIntr(object). Availability of an interaction depends on the LRscore slot to be used as well as the significance metric to be used.
slot.use

The LRscore type to be used. See vignette for explanation.
signif.use

The significance metric to be used. See vignette for explanation.
num.per.choose

The maximum number of genes to be chosen for each wilcoxon test. A test happens for each interaction and each cell type. Default 50.
alpha.test

A sequence of alpha values to be tested for the elastic net regression model. The best alpha value will be chosen based on the smallest loss. Larger alpha values result in less final selection. Default seq(0.5, 1, 0.1).
seed

Random seed for controlling the cross validation. Default 1.
minCell

Minimum number of cells where a gene is expressed to be considered. Default 50.
verbose

Whether to print progress messages. Default TRUE.

Value

list object, each element is the result for each interaction. Each interaction result is a list object containing the following entries:

  • glmnet_model - A "cv.glmnet" object, the regression model.

  • score_refine - A 1-column matrix of refined LR scores of this interaction in each cell.

  • coef - A matrix of coefficients of the regression model.

  • sign_clusters - A character vector of significant clusters in selected by the model.

  • sign_genes - A character vector of significant genes selected by the model.

  • alpha - The alpha value used for the final chosen model.

  • slot.use - The LRScore type used for this analysis.

  • signif.use - The significance metric used for this analysis.

Calculate LRScore from the imputed L and R values

Description

After running imputeLR, we can calculate the LR score for each location. The LR score is calculated as the product of the imputed LL and RR values. With the LR score inferred, we subsequently perform permutation tests to construct the null distribution for testing the significance of the interactions.

Usage

inferIntrScore(
  object,
  recep.smooth = FALSE,
  intr.type = c("diff", "cont"),
  perm.size = 1e+05,
  norm.method = c("default", "cpm", "none", "scanpy"),
  numCores = 1
)

Arguments

object

A CytoSignal object, with imputeLR already run.
recep.smooth

A logical scalar. Whether to use the smoothed RR values which is imputed with DT method. Default FALSE.
intr.type

A character vector. The type of interactions to calculate the LR score. Choose from one or both of "diff" and "cont" for diffusion-dependent and contact-dependent interactions, respectively. Default uses both.
perm.size

A numeric scalar. The number of permutations to perform. Default 1e5 times.
norm.method

The normalization method to apply to the imputed L and R
numCores

SPARK::sparkx parameter. The number of cores to use. Default 1.

Value

A CytoSignal object updated. Entries in object@lrscore slot will be updated with the LR scores and the significance inferrence. When recep.smooth is by default FALSE, the object@lrscore slot will be updated with object@lrscore$`GauEps-Raw` and object@lrscore$`DT-Raw`. When recep.smooth is TRUE, the object@lrscore slot will be updated with object@lrscore$`GauEps-DT` and object@lrscore$`DT-DT`.

Examples

## Not run: 
object <- findNN(object)
object <- imputeLR(object)
object <- inferIntrScore(object)

## End(Not run)

Calculate the interaction velocity from the imputed time derivative of LL or RR values

Description

After running imputeVeloLR, we can calculate the interaction velocity for each location. Please refer to the manualscript for detail of the calculation.

Usage

inferIntrVelo(object, recep.smooth = FALSE, norm.method = "scanpy")

Arguments

object

A CytoSignal object, with imputeVeloLR already run.
recep.smooth

A logical scalar. Whether to use the smoothed RR values which is imputed with DT method. Default FALSE.
norm.method

The normalization method to apply to the velocity data, need to be consistent with the normalization method used when generating the input RNA velocity. Please consult the tool used for it, e.g. scVelo, VeloVAE. Default is "scanpy". Can choose from "scanpy", "cpm", "default" or "none".

Value

A CytoSignal object updated. Entries in object@lrvelo slot will be updated with the velocity scores. When recep.smooth is by default FALSE, the object@lrvelo slot will be updated with object@lrvelo$`GauEps-Raw` and object@lrvelo$`DT-Raw`. When recep.smooth is TRUE, the object@lrvelo slot will be updated with object@lrvelo$`GauEps-DT` and object@lrvelo$`DT-DT`.

Examples

## Not run: 
object <- addVelo(object, velo.s, velo.u)
object <- findNN(object)
object <- imputeVeloLR(object)
object <- inferIntrVelo(object)

## End(Not run)

Permute LR score for specific ligand-receptor imputation obj pairs

Description

This function is a follow-up function of inferScoreLR. It computes the NULL LR-scores using the NULL imputation results and stores the results in the LR score object. The null distribution of the LR scores can be used to test the significance of the LR scores.

Usage

inferNullScoreLR(object, slot.use = NULL)

Arguments

object

A Cytosignal object
slot.use

Which LR score to use. Use the name specified with tag when running inferLRScore.

Value

A Cytosignal object

Compute the LR score for specific ligand-receptor imputation pairs

Description

Compute the LR score for specific ligand-receptor imputation pairs

Usage

inferScoreLR(
  object,
  lig.slot,
  recep.slot,
  norm.method = c("default", "cpm", "none", "scanpy"),
  intr.db.name = c("diff_dep", "cont_dep"),
  tag = paste0(lig.slot, "-", recep.slot)
)

Arguments

object

A Cytosignal object
lig.slot

The ligand slot to use
recep.slot

The receptor slot to use
norm.method

Method to normalize the data. Default is "default".
intr.db.name

The intr database name to use
tag

Name of the result to be stored in object.

Value

A Cytosignal object

Infer significance of LR scores

Description

Infer significance of LR scores

Usage

inferSignif(
  object,
  fdr.method = c("spatialFDR", "fdr"),
  p.value = 0.05,
  reads.thresh = 100,
  sig.thresh = 100,
  slot.use = NULL,
  nn.use = NULL
)

Arguments

object

A Cytosignal object
fdr.method

The false discovery rate method to use. Choose from "spatialFDR" and "fdr". Default "spatialFDR".
p.value

A numeric scalar. The p-value threshold to use for filtering significant interactions. Default 0.05.
reads.thresh

A numeric scalar. The minimum number of reads for a ligand-receptor interaction to be considered. Default 100.
sig.thresh

A numeric scalar. The minimum number of of beads for a ligand-receptor interaction to be considered. Default 100.
slot.use

Which LR score to use. Use the name specified with tag when running inferIntrScore. Default NULL apply specified significance and filtering criteria to all available LRscore slots.
nn.use

Which imputation to use. Default the imputation used for deriving the LRScore specified with slot.use. Use the name specified with tag when running findNNGauEB; use "DT" for imputation produced with findNNDT; or use "Raw" for imputation produced with findNNRaw.

Value

A Cytosignal object

Infer the correspondence between LR-scores and Significance

Description

Use DT neighbors to impute the p-value and LR-score for each cell, then compute the pearson correlation between the imputed LR-score and p-value.

Usage

inferSpatialCorr(object, correctBy = c("cell", "intr"), slot.use = NULL)

Arguments

object

A Cytosignal object
correctBy

Spatial FDR correction method. Choose from "cell" or "intr".
slot.use

Which LR score to use. Use the name specified with tag when running inferLRScore.

Value

A Cytosignal object

Sub function for inferVeloLR, input a CytoSignal object

Description

Sub function for inferVeloLR, input a CytoSignal object

Usage

inferVeloLR(
  object,
  lig.slot,
  recep.slot,
  intr.db.name,
  norm.method = "scanpy",
  tag = NULL
)

Arguments

object

A Cytosignal object
lig.slot

The ligand slot to use
recep.slot

The receptor slot to use
intr.db.name

The intr database name to use
norm.method

The normalization method to apply to the counts, need to be consistent with the normalization method used for the RNA velocity method. Default is "scanpy".
nn.use

slot that the neighbor index should be taken from, by default is the same as the recep.slot. For example, if velo.obj = GauEps-DT, then nn.use = "DT". nn.use could also be a user-defind factor.

Value

A Cytosignal object

The lrScores Class

Description

The lrScores object is created from one ST dataset. User could choose two imputation methods to calculate the ligand-receptor scores. The class stores the ligand, receptor, and interaction database, the ligand-receptor scores, the ligand-receptor scores for permuted data, the ligand-receptor scores for permuted data, and the log of each main steps.

Details

The key slots used in the lrScores object are described below.

Slots

lig.slot

ligand database

recep.slot

receptor database

intr.slot

interaction database

intr.list

list of interaction database

score

ligand-receptor scores

score.null

permuted ligand-receptor scores

res.list

list of results

log

Log of each main steps

The lrvelo Class

Description

The lrvelo object is created from one ST dataset. User could choose two imputation methods to calculate the ligand-receptor scores. The class stores the index of ligand, receptor, and interaction database, inferred lrvelo for each interaction, and the log of each main steps.

Details

The key slots used in the lrVelo object are described below.

Slots

lig.slot

ligand database

recep.slot

receptor database

intr.slot

interaction database

intr.list

list of interaction database

velo.s

A matrix of spliced velo for each gene

velo.u

A matrix of unspliced velo for each gene

velo.intr

A sparse matrix of velo for each intr

nn.id

A factor of nearest neighbor id. Re-do findNN since the order of cells may change!

nn.dist

A factor of nearest neighbor distance. Re-do findNN since the order of cells may change!

log

Log of each main steps

Permute Imputation Results of specific imputation method

Description

This function permutes the imputation methods from which a given LR score is calculated. Note that all rounds of permutation will use the same shuffle and sample index.

Usage

permuteLR(object, slot.use = NULL, norm.method = "default", perm.size = 1e+05)

Arguments

object

A CytoSignal object with inferLRScore run beforehand.
slot.use

Which LR score to use. Use the name specified with tag when running inferLRScore.
perm.size

Size of the permutation test. Defult 100000.
norm.me

Normalization method. See normCounts for detail.

Value

A CytoSignal object

Plotting edge for a given interaction from a CytoSignal object

Description

Plotting edge for a given interaction from a CytoSignal object

Usage

plotEdge(
  object,
  intr,
  type = c("receiver", "sender"),
  slot.use = NULL,
  signif.use = NULL,
  colors.list = NULL,
  return.plot = TRUE,
  plot_dir = "csEdgePlot/",
  filename = NULL,
  plot.fmt = c("png", "pdf", "svg"),
  title = NULL,
  edge.size = 500,
  use.shape = 16,
  line.width = 0.01,
  use.phi = 30,
  use.theta = -17,
  z.scaler = 0.03,
  box = TRUE,
  z.pt.interval = 1,
  pt.size = 0.1,
  pt.sig.size = NULL,
  pt.stroke = 0.2,
  width = 5,
  height = 5,
  set.res = 300,
  verbose = TRUE
)

Arguments

object

CytoSignal object.
intr

Interaction to plot. See available options with showIntr.
type

Type of plot, either "sender" or "receiver". Default "sender".
slot.use

Slot to use for plotting
signif.use

Significance level to use for plotting
colors.list

List of colors to use for plotting
return.plot

Whether to return "plist" object for figure organization. TRUE returns "plist" which can be shown on current display device with plot(). FALSE saves the figure to disk. Default TRUE.
plot_dir

Path where the figure will be saved when return.plot = FALSE. Default "csEdgePlot/" (under current working directory).
filename

Filename of the figure inside plot_dir. Please match extension name with plot.fmt and do not include path. Default NULL and the exact filename will be determined by interaction information and plot.fmt.
plot.fmt

Format of plot, either "png" or "pdf".
title

Title of plot
use.shape

Shape of points
line.width

Width of lines
use.phi

Angle of 3D plot
use.theta

Angle of 3D plot
z.scaler

Scaling factor for z-axis
z.pt.interval

Interval of points on z-axis
pt.stroke

Width of points
set.res

Resolution of plot
use.cex

Size of points
u_width

Width of plot
u_hgt

Height of plot

Value

plist object when return.plot = TRUE, no in memory object is returned when return.plot = FALSE but the figure will be saved on disk.

Plot the refined score of each interaction after regression model refinement

Description

Plot the refined score of each interaction after regression model refinement

Usage

plotRefinedScore(
  object,
  intrDEGRes,
  intr = NULL,
  pt.size = 0.5,
  pt.stroke = 0.1
)

Arguments

object

A Cytosignal object with inferIntrScore already run.
intrDEGRes

The direct output object of inferIntrDEG.
intr

A vector of unique interaction IDs that are available in intrDEGRes, or numerical index within its range. Default NULL use all the results.
pt.size

Size of the points in the plot. Default 0.5.
pt.stroke

Stroke size of the points in the plot. Default 0.1.

Value

List of ggplot objects, each shows the refined LR score of each selected interaction.

Create scatter plot from REVIGO result

Description

Create scatter plot from REVIGO result

Usage

plotREVIGO(
  rvg,
  labelBy = c("Name", "TermID"),
  labelIdx = NULL,
  labelSize = 4,
  max.overlaps = 5
)

Arguments

rvg

A data.frame from revigo. Must contain columns: PC_0, PC_1, LogSize, Frequency, and Name.
labelBy

From which field of the revigo result to retrieve label text. Choose from "Name" or "TermID". Default "Name".
labelIdx

Index matching to rows of rvg to indicate whether a label has to be shown for the bubble. Default NULL automatically shows tidy view with ggrepel.
labelSize

Size of text label for marked dots. Default 4.
max.overlaps

?ggrepel::geom_text_repel parameter. Default 5.

Value

A ggplot object

Examples

rvg <- revigo(c("GO:0003002", "GO:0007389", "GO:0007423", "GO:0009653"),
              speciesTaxon = 10090)
plotREVIGO(rvg)

Plot edges by each cluster as sender or receiver cells

Description

By default ranked by the number of significant cells in each cluster

Usage

plotSigCluster(
  object,
  plot_dir,
  cluster.list = NULL,
  intr.num = 10,
  type = c("sender", "receiver"),
  slot.use = NULL,
  signif.use = NULL,
  colors.list = NULL,
  plot.fmt = "png",
  edge.size = 2000,
  all.in.one = T,
  plot.all.sig = F,
  use.cex = 0.1,
  use.shape = 16,
  line.width = 0.02,
  use.phi = 30,
  use.theta = -17,
  z.scaler = 0.03,
  z.pt.interval = 1,
  pt.stroke = 0.2,
  u_width = 6,
  u_hgt = 6,
  set.res = 400,
  return.plot = F
)

Arguments

object

A CytoSignal object
plot_dir

Directory to save plots
cluster.list

A list of clusters to plot
intr.num

Number of interactions to plot
type

Either "sender" or "receiver"
slot.use

Slot to use for plotting
signif.use

Significance threshold to use for plotting
colors.list

A list of colors to use for each cluster
plot.fmt

Plot format
edge.size

Number of edges to plot
all.in.one

Plot all clusters in one plot
plot.all.sig

Plot all significant edges
use.shape

Shape of points
line.width

Width of lines
use.phi

Angle of view
use.theta

Angle of view
z.scaler

Scale of z-axis
z.pt.interval

Interval of z-axis
pt.stroke

Stroke of points
u_width

Width of plot
u_hgt

Height of plot
set.res

Resolution of plot
return.plot

Return plot object
pt.size

Size of points
...

Other arguments

Value

A list of plots

Plot significant interactions ranked by the user-specified metric

Description

Plot significant interactions ranked by the user-specified metric

Usage

plotSignif(
  object,
  num.plot = NULL,
  res_dir,
  plot.details = T,
  slot.use = NULL,
  signif.use = NULL,
  plot.clusters = T,
  plot.velo = F,
  colors.list = NULL,
  pt.size = 0.1,
  pt.stroke = 0.2,
  u_width = 6,
  u_hgt = 5,
  set.res = 200,
  return.plot = F
)

Arguments

object

A cytosignal object
num.plot

Number of interactions to plot
res_dir

Directory to save the plots
plot.details

Whether to plot NULL imputed values and scores
slot.use

The LRscore slot to use for plotting
signif.use

The metric used to rank the interactions, by default "result.hq.pear"
plot.clusters

Whether to plot the clusters
plot.velo

Whether to plot the velocity
colors.list

A list of colors to use for plotting
pt.size

Size of the points
pt.stroke

Stroke of the points
u_width

Width of the plot
u_hgt

Height of the plot
set.res

Resolution of the plot
return.plot

Whether to return the plot

Value

A plot if return.plot is TRUE. Otherwise, plots are saved to the specified directory.

Plot significant interactions ranked by the user-specified metric

Description

Plot significant interactions ranked by the user-specified metric

Usage

plotSignif2(
  object,
  intr,
  edge = FALSE,
  velo = FALSE,
  slot.use = NULL,
  signif.use = NULL,
  colors.list = NULL,
  pt.size = 0.1,
  pt.stroke = 0.2,
  return.plot = FALSE,
  plot_dir = "csSignifPlot/",
  plot.fmt = c("png", "pdf", "svg"),
  raster = NULL,
  resolution = 300,
  verbose = FALSE
)

Arguments

object

A cytosignal object
intr

Specify interactions to be plotted. A vector of either the unique ID of interactions or the numeric indices. Available IDs can be shown with showIntr(object). Availability of an interaction depends on the LRscore slot to be used as well as the significance metric to be used.
edge, velo

Logical, whether to plot edge or velocity, respectively.
slot.use

The LRscore slot to use for plotting
signif.use

The metric used to rank the interactions, by default "result.hq.pear"
colors.list

A list of colors to use for plotting
pt.size

Size of the points
pt.stroke

Stroke of the points
return.plot

Whether to return the plot
plot_dir

Directory to save the plots
plot.fmt

Format of output file. "png", "pdf", or "svg".
resolution

Resolution of the output figure.
verbose

Logical, whether to show progress.

Value

A plot if return.plot is TRUE. Otherwise, plots are saved to the specified directory.

Plot 3D LR-velo ranked by the user-specified metric

Description

Plot 3D LR-velo ranked by the user-specified metric

Usage

plotVelo(
  object,
  intr,
  return.plot = TRUE,
  plot_dir = "csVeloPlot/",
  filename = NULL,
  plot.fmt = c("png", "pdf"),
  slot.use = NULL,
  signif.use = NULL,
  use.clusters = NULL,
  colors.list = NULL,
  z.scaler = 0.03,
  title = NULL,
  pt.size = 0.1,
  use.shape = 16,
  use_xbins = 15,
  use_ybins = 15,
  arrow.line.width = 0.6,
  arrow.width = 0.06,
  pt.stroke = 0.2,
  use.phi = 30,
  use.theta = -17,
  box = TRUE,
  axis.arrow.len = 1,
  width = 6,
  height = 6,
  set.res = 300,
  verbose = TRUE
)

Arguments

object

A cytosignal object
return.plot

Whether to return the plot
slot.use

The LRscore slot to use for plotting
signif.use

The metric used to rank the interactions, by default "result.hq.pear"
use.clusters

Plot only selected clusters. Default NULL for all clusters.
colors.list

A list of colors to use for plotting
z.scaler

Scaling factor for the z-axis
pt.size

Point size. Default 0.1
use.shape

Point shape. Default 16
use_xbins

Number of bins for the x-axis. Default 15
use_ybins

Number of bins for the y-axis. Default 15
arrow.line.width

Width of the arrow line. Default 0.6
arrow.width

Width of the arrow. Default 0.06
pt.stroke

Stroke of the points
use.phi

Set view angel: phi the colatitude. Default 30
use.theta

Set view angel: theta gives the azimuthal direction. Default -17
box

Whether to show a box panel for the 3D region. Default TRUE
axis.arrow.len

When box = FALSE, set the length of the axis arrows.
width

Width of the plot. Default 6
height

Height of the plot. Default 6
set.res

Resolution of the plot. Default 300.
num.plot

Number of interactions to plot
res_dir

Directory to save the plots
plot.velo

Whether to plot the velocity

Value

A plot if return.plot is TRUE. Otherwise, plots are saved to the specified directory.

Print the CytosignalIntrDEG object representation to screen

Description

Print the CytosignalIntrDEG object representation to screen

Usage

## S3 method for class 'CytosignalIntrDEG'
print(x, ...)

Arguments

x

A CytosignalIntrDEG object.
...

Passed to other methods

Remove imputed data and normalized imputed data from CytoSignal object to save disk space

Description

Remove imputed data and normalized imputed data from CytoSignal object to save disk space

Usage

purgeBeforeSave(object, purge.raw = TRUE, purge.null = FALSE)

Arguments

object

CytoSignal object

Value

a CytoSignal object

Rank the inferred high-quality interactions by their spatial variability

Description

This function utilizes SPARK package to calculate the spatial variability of the high-quality interactions, using their LR scores. Please refer to Jiaqiang Zhu, et al., 2021, Genome Biology for more details of the method.

Usage

rankIntrSpatialVar(object, slot.use = NULL, numCores = 1, verbose = FALSE)

Arguments

object

A CytoSignal object, with inferSignif already run.
slot.use

Which LR score to use. Use the name specified with tag when running inferLRScore.
numCores

SPARK::sparkx parameter, an integer specifying the number of threads.
verbose

SPARK::sparkx parameter, a logical value indicating whether to print details for debug purpose

Value

The input CytoSignal object with the spatially variable high-quality interaction list updated at object@lrscore[[slot.use]]@res.list$result.spx

Examples

## Not run: 
object <- findNN(object)
object <- imputeLR(object)
object <- inferScoreLR(object, lig.slot = "GauEps", recep.slot = "Raw",
                       intr.db.name = "diff_dep")
object <- permuteLR(object)
object <- inferNullScoreLR(object)
object <- inferSignif(object)
object <- rankIntrSpatialVar(object)

## End(Not run)

Remove low quality cells and genes from raw counts

Description

Remove low quality cells and genes from raw counts

Usage

removeLowQuality(object, counts.thresh = 300, gene.thresh = 50)

Arguments

object

CytoSignal object
counts.thresh

threshold for cell counts
gene.thresh

threshold for gene counts

Value

a CytoSignal object

Submit query to REVIGO and get result to R

Description

This function gets an input of vector of GO terms and submit query to REVIGO via HTTP, and retrieves the result from the job. The code is based on example provided by REVIGO, originally located at http://revigo.irb.hr/FAQ

Please visit http://revigo.irb.hr/ for the web-based interactive tool, and please see https://doi.org/10.1371/journal.pone.0021800 for detailed introduction to the functionality.

Usage

revigo(
  GOterms,
  cutoff = 0.7,
  value = NULL,
  valueType = c("PValue", "Higher", "Lower", "HigherAbsolute", "HigherAbsLog2"),
  speciesTaxon = "0",
  measure = c("SIMREL", "LIN", "RESNIK", "JIANG"),
  removeObsolete = TRUE
)

Arguments

GOterms

Character vector of GO term IDs.
cutoff

Size of the result, between 0.4 to 0.9. Larger value returns more result. Default 0.7.
value

A vector of additional information for each GO term. Default NULL does not include additional information. Type of the value has to be specified with valueType.
valueType

Type of value. Default "PValue". Optionally, "Higher", "Lower", "HigherAbsolute", "HigherAbsLog2".
speciesTaxon

Taxon ID of the species. Default "0" queries to all available species. "10090" is for mouse and "9606" is for human.
measure

The semantic similarity measure to use. Default "SIMREL". Optionally, "LIN", "RESNIK", "JIANG".
removeObsolete

Logical. Whether to remove obsolete GO terms. Default TRUE.

Value

A data.frame with many information

Examples

rvg <- revigo(c("GO:0003002", "GO:0007389", "GO:0007423", "GO:0009653"),
              speciesTaxon = 10090)

Identify spatially significant interactions using std-corrected pearson correlation Normal Moran's I test is not applicable here since the total number of the cell is too large, causing unnacceptable computation cost. Here we use a modified version of Moran's I test, which is to take only the top KNNs to compute the Moran's I test.

Description

Identify spatially significant interactions using std-corrected pearson correlation Normal Moran's I test is not applicable here since the total number of the cell is too large, causing unnacceptable computation cost. Here we use a modified version of Moran's I test, which is to take only the top KNNs to compute the Moran's I test.

Usage

runPears.std(object, k = 10, weight = 2, score.slot = NULL)

Arguments

object

A Cytosignal object
k

The number of nearest neighbors to use
weight

The weight of the nearest neighbors
score.slot

Which LR score to use. Use the name specified with tag when running inferLRScore.

Value

A Cytosignal object

show method for CytoSignal

Description

show method for CytoSignal

Usage

## S4 method for signature 'CytoSignal'
show(object)

Arguments

object

CytoSignal object

show method for cytosignal obj

Description

show method for cytosignal obj

Usage

## S4 method for signature 'ImpData'
show(object)

show method for ImpData

Description

show method for ImpData

Usage

showImp(object, slot.use = NULL)

Arguments

object

CytoSignal object
slot.use

slot to use

show method for CytoSignal

Description

show method for CytoSignal

Usage

showIntr(object, slot.use = NULL, signif.use = NULL, return.name = FALSE)

Arguments

object

CytoSignal object
slot.use

slot to use
signif.use

Significance level to use for plotting
return.name

Whether to return interaction name that comes with the form of "ligand-receptor", instead of showing the unique IDs. Default FALSE.

Value

By default, character vector of unique IDs of the interactions available to the specified slots. When return.name = TRUE, a named character vector where the names are the IDs and the values are interaction names with the form of "ligand-receptor".

show all current logs

Description

show all current logs

Usage

showLog(object)

Arguments

object

CytoSignal object

show method for lrScores

Description

show method for lrScores

Usage

showScore(object, slot.use = NULL)

Arguments

object

CytoSignal object
slot.use

slot to use

show intr.data in ImpData

Description

show intr.data in ImpData

Usage

showUnpt(object, slot.use = NULL)

Arguments

object

CytoSignal object
slot.use

slot to use

show method for lrVelo

Description

show method for lrVelo

Usage

showVelo(object, slot.use = NULL)

Arguments

object

CytoSignal object
slot.use

slot to use

Suggest scaling factor of real units to spatial units

Description

This function is used to estimate the scaling facor of real units to spatial units. Returns the top 5 possible scaling factors, users can choose to which to use.

Usage

suggestScaleFactor(object, cell.intervel = NULL)

Arguments

object

CytoSignal object
cell.intervel

Distance between two cell in physical units

Value

a vector of possible scaling factors

Generate character vector of unique color hex codes Base color palette adopts ggsci:::ggsci_db$igv$default

Description

Generate character vector of unique color hex codes Base color palette adopts ggsci:::ggsci_db$igv$default

Usage

uniqueColors(n)

Arguments

n

Number of unique colors to generate. When not exceeding existing number of colors, just return the first n colors, otherwise generate interpolated colors.