| Title: | Predict genes expressed selectively in specific cell types |
|---|---|
| Description: | Infers cell type-specific expression based on co-expression similarity with known cell type marker genes. Can make accurate predictions using publicly available expression data, even when a cell type has not been isolated before. |
| Authors: | Brad Nelms |
| Maintainer: | Brad Nelms <[email protected]> |
| License: | Artistic-2.0 |
| Version: | 1.31.0 |
| Built: | 2024-09-18 05:12:15 UTC |
| Source: | https://github.com/bioc/CellMapper |
Container for expression data that has been pre-processed using the 'CMprep' function. A 'CellMapperList' object can be provided directly to the 'CMsearch' function to predict genes expressed selectively in specific cell types.
'CMprep' transforms an expression matrix using singular value decomposition (SVD), resulting in a matrix, 'B', with the left-singular vectors of original data matrix and a vector, 'd', with the singular values. Singular vectors that account for less variance than an individual sample in the original dataset have been trimed (Kaiser's criterion), thereby removing singular vectors that mainly account for noise. The advantage of this transformation is that it reduces dataset size, and avoids the need to perform a time-consuming SVD transformation before running 'CMsearch'.
A 'CellMapperList' object contains the transformed left singular value matrix, 'B', and singular values, 'd', as well as meta-data associated with the original expression matrix.
'CellMapperList' instances are usually created through the 'CMprep' function. See ?CMprep
To create a 'CellMapperList' object directly, the following constructor is provided:
CellMapperList(B, d, meta = list())
where B is a numeric matrix containing left-singular vectors, d is a numeric vector containing singular values, and meta is a list object containing metadata.
Same as for SimpleList objects. See ?SimpleList
The sample metadata can be extracted using metadata().
Brad Nelms
B.D. Nelms, L. Waldron, L.A. Barrera, A.W. Weflen, J.A. Goettel, G. Guo, R.K. Montgomery, M.R. Neutra, D.T. Breault, S.B. Snapper, S.H. Orkin, M.L. Bulyk, C. Huttenhower and W.I. Lencer. CellMapper: rapid and accurate inference of gene expression in difficult-to-isolate cell types. Genome Biology 2016, 17(1).
# Create a mock expression dataset with random expression values ngenes <- 1000 narrays <- 100 x <- matrix(rnorm(ngenes*narrays), ngenes, narrays) rownames(x) <- 1:ngenes # Prepare a CellMapperList object using the CMprep function data <- CMprep(x, DataSource = "Mock Expression Matrix") show(data) metadata(data)# Create a mock expression dataset with random expression values ngenes <- 1000 narrays <- 100 x <- matrix(rnorm(ngenes*narrays), ngenes, narrays) rownames(x) <- 1:ngenes # Prepare a CellMapperList object using the CMprep function data <- CMprep(x, DataSource = "Mock Expression Matrix") show(data) metadata(data)
Prepares a dataset for use with the CMsearch function.
CMprep(Data, DataSource = '', GeneIDType = '', verbose = TRUE)CMprep(Data, DataSource = '', GeneIDType = '', verbose = TRUE)
Data |
a numeric |
DataSource |
An optional character vector of length one providing some information about the expression data source. |
GeneIDType |
An optional character vector of length one that lists the type of gene IDs used in the expression dataset (e.g. "Human Entrez IDs"). |
verbose |
logical value indicating whether progress updates should be provided. |
This function calculates the singular value decomposition of a gene expression
matrix and prepares the data for use with the CMsearch function. It can
take some time for large expression matrices, and so it is recommended to save
the output for future use. Pre-processed microarray data, ready for immediate
use with CMsearch, can be found in the CellMapperData package.
See the CellMapper vignette for examples about how to use CMprep and
associated functions to infer genes selectively expressed in specific cell
types.
A 'CellMapperList' object ready to be provided to the CMsearch function.
Brad Nelms
B.D. Nelms, L. Waldron, L.A. Barrera, A.W. Weflen, J.A. Goettel, G. Guo, R.K. Montgomery, M.R. Neutra, D.T. Breault, S.B. Snapper, S.H. Orkin, M.L. Bulyk, C. Huttenhower and W.I. Lencer. CellMapper: rapid and accurate inference of gene expression in difficult-to-isolate cell types. Genome Biology 2016, 17(1).
CMsearch, ExpressionSet,
CellMapperData
# Create a mock expression dataset with random expression values ngenes <- 1000 narrays <- 100 x <- matrix(rnorm(ngenes*narrays), ngenes, narrays) rownames(x) <- 1:ngenes # Prepare the dataset for use with CMsearch data <- CMprep(x) show(data) # Save the processed dataset for later ## Not run: save(data, file = "Preprocessed_CellMapper_Data.RData") ## End(Not run)# Create a mock expression dataset with random expression values ngenes <- 1000 narrays <- 100 x <- matrix(rnorm(ngenes*narrays), ngenes, narrays) rownames(x) <- 1:ngenes # Prepare the dataset for use with CMsearch data <- CMprep(x) show(data) # Save the processed dataset for later ## Not run: save(data, file = "Preprocessed_CellMapper_Data.RData") ## End(Not run)
Predicts which genes are selectively expressed in the same cell type as a given cell type-specific marker gene (the 'query gene'), based on co-expression similarity.
CMsearch(Data, query.genes = NULL, control.genes = NULL, QDW = TRUE, alpha = 0.5, verbose = TRUE, raw.pvals = FALSE)CMsearch(Data, query.genes = NULL, control.genes = NULL, QDW = TRUE, alpha = 0.5, verbose = TRUE, raw.pvals = FALSE)
Data |
a |
query.genes |
a list of genes that are specifically expressed in the cell type of interested, supplied as a character vector of gene names (matching the row names of the original expression matrix). |
control.genes |
a list of genes expressed specifically in non-target cell types (optional), supplied as a character vector of gene names (matching the row names of the original expression matrix). This option generally has little effect on the results and its use is NOT recommended. |
QDW |
logical value indicating whether 'query driven weighting' should be applied in the CellMapper SVD filter. The default value of TRUE can be used in most cases. |
alpha |
alpha parameter controlling the strength of the CellMapper SVD filter. The default value of 0.5 can be used in most cases. |
verbose |
logical value indicating whether progress updates should be provided. |
raw.pvals |
logical value indicating whether unadjusted p-values, which have not been corrected for multiple hypothesis testing, should be returned. |
This function predicts which genes are selectively expressed in the same cell
type as a given cell type-specific marker gene (the 'query gene'), based on
co-expression similarity. The only required inputs are a gene expression matrix
that has been pre-processed with the CMprep function (or a list of
pre-processed expression matrices), and a 'query gene' known to be specifically
expressed in the cell type of interest. Pre-processed microarray data, ready for
immediate use with CMsearch, can be found in the CellMapperData
package.
See the CellMapper vignette for examples about how to use CMsearch and
associated functions to infer genes selectively expressed in specific cell
types.
A dataframe containing the gene identifiers in the first column, the false discovery rate (FDR) in the second, and the unadjusted p-value in the third. FDR is calculated using the method of Benjamini and Hochberg.
Brad Nelms
B.D. Nelms, L. Waldron, L.A. Barrera, A.W. Weflen, J.A. Goettel, G. Guo, R.K. Montgomery, M.R. Neutra, D.T. Breault, S.B. Snapper, S.H. Orkin, M.L. Bulyk, C. Huttenhower and W.I. Lencer. CellMapper: rapid and accurate inference of gene expression in difficult-to-isolate cell types. Genome Biology 2016, 17(1).
# Create a mock expression dataset with random expression values ngenes <- 1000 narrays <- 100 x <- matrix(rnorm(ngenes*narrays), ngenes, narrays) rownames(x) <- 1:ngenes # Prepare the dataset for use with CMsearch data <- CMprep(x) # Predict which genes are co-expressed in the same cell type as 'gene' 59 in the # mock expression dataset results <- CMsearch(data, query.genes = '59') head(results)# Create a mock expression dataset with random expression values ngenes <- 1000 narrays <- 100 x <- matrix(rnorm(ngenes*narrays), ngenes, narrays) rownames(x) <- 1:ngenes # Prepare the dataset for use with CMsearch data <- CMprep(x) # Predict which genes are co-expressed in the same cell type as 'gene' 59 in the # mock expression dataset results <- CMsearch(data, query.genes = '59') head(results)
Query genes, positive control genes, and negative control genes for example cell types.
data(ExampleQueryGenes) data(PositiveControlGenes) data(NegativeControlGenes)data(ExampleQueryGenes) data(PositiveControlGenes) data(NegativeControlGenes)
'ExampleQueryGenes' is a data frame listing the cell types and query genes
analyzed in Nelms, et al. (2016). These genes can be used as query or control
genes for custom CellMapper searches, and are also provided for reproducibility
of the manuscript results. All predictions from the manuscript can be generated
using the ReplicateManuscript function. The 'Dataset' column indicates
which dataset(s) were used for each cell type, available from Bioconductor's
ExperimentHub. All datasets are available from the CellMapperData
package.
'PositiveControlGenes' is a list of positive control genes for each cell type used in Nelms, et al. (2016). These positive control genes were NOT provided to the algorithm, but were used to assess the accuracy of the results. This list contains Entrez gene IDs.
'NegativeControlGenes' is a list of negative control genes for each cell type used in Nelms, et al. (2016). These negative control genes were NOT provided to the algorithm, but were used to assess the accuracy of the results. This list contains Entrez gene IDs.
'ExampleQueryGenes' is a data frame, 'PositiveControlGenes' and 'NegativeControlGenes' are both lists of character vectors
Brad Nelms
B.D. Nelms, L. Waldron, L.A. Barrera, A.W. Weflen, J.A. Goettel, G. Guo, R.K. Montgomery, M.R. Neutra, D.T. Breault, S.B. Snapper, S.H. Orkin, M.L. Bulyk, C. Huttenhower and W.I. Lencer. CellMapper: rapid and accurate inference of gene expression in difficult-to-isolate cell types. Genome Biology 2016, 17(1).
CMsearch, ReplicateManuscript,
CellMapperData,
ExperimentHub
data(ExampleQueryGenes) head(ExampleQueryGenes)data(ExampleQueryGenes) head(ExampleQueryGenes)
Recreates the CellMapper predictions from Nelms, et al. (2016) – Additional dataset 1.
ReplicateManuscript()ReplicateManuscript()
This function will recreate the CellMapper predictions from Nelms, et al.
(2016), Additional dataset 1. It requires the ExperimentHub and
CellMapperData packages to be installed. Provided that the
CellMapperData resources have previously been downloaded from
ExperimentHub, the 'ReplicateManuscript' function should take ~5 minutes
to run on a personal laptop.
A 'list' object containing CellMapper predictions for each cell type. Genes names are returned as Entrez identifiers.
Brad Nelms
B.D. Nelms, L. Waldron, L.A. Barrera, A.W. Weflen, J.A. Goettel, G. Guo, R.K. Montgomery, M.R. Neutra, D.T. Breault, S.B. Snapper, S.H. Orkin, M.L. Bulyk, C. Huttenhower and W.I. Lencer. CellMapper: rapid and accurate inference of gene expression in difficult-to-isolate cell types. Genome Biology 2016, 17(1).
CMsearch, ExampleQueryGenes,
CellMapperData,
ExperimentHub
Results <- ReplicateManuscript() head(Results[['GABAergic Neurons']])Results <- ReplicateManuscript() head(Results[['GABAergic Neurons']])