progeny: Calculate PROGENy pathway scores from gene expression
In progeny: Pathway RespOnsive GENes for activity inference from gene expression

Description Usage Arguments Details Value See Also Examples

This function uses the linear model of pathway-responsive genes underlying the PROGENy method. It transforms a gene expression matrix with HGNC/MGI gene symbols in rows and sample names in columns into a pathway score matrix with samples in rows and pathways in columns.

progeny(
  expr,
  scale = TRUE,
  organism = "Human",
  top = 100,
  perm = 1,
  verbose = FALSE,
  z_scores = FALSE,
  get_nulldist = FALSE,
  assay_name = "RNA",
  return_assay = FALSE,
  ...
)

`expr`	A gene expression object with HGNC/MGI symbols in rows and samples in columns. In order to run PROGENy in single-cell RNAseq data, it also accepts Seurat and SingleCellExperiment object, taking the normalized counts for the computation.
`scale`	A logical value indicating whether to scale the scores of each pathway to have a mean of zero and a standard deviation of one. It does not apply if we use permutations.
`organism`	The model organism - "Human" or "Mouse"
`top`	The top n genes for generating the model matrix according to significance (p-value)
`perm`	An interger detailing the number of permutations. No permutations by default (1). When Permutations larger than 1, we compute progeny pathway scores and assesses their significance using a gene sampling-based permutation strategy, for a series of experimental samples/contrasts.
`verbose`	A logical value indicating whether to display a message about the number of genes used per pathway to compute progeny scores (i.e. number of genes present in the progeny model and in the expression dataset)
`z_scores`	Only applies if the number of permutations is greater than 1. A logical value. TRUE: the z-scores will be returned for the pathway activity estimations. FALSE: the function returns a normalized z-score value between -1 and 1.
`get_nulldist`	Only applies if the number of permutations is greater than 1. A logical value. TRUE: the null distributions generated to assess the signifance of the pathways scores is also returned.
`assay_name`	Only applies if the input is a Seurat object. It selects the name of the assay on which Progeny will be run. Default to: RNA, i.e. normalized expression values.
`return_assay`	Only applies if the input is a Seurat object. A logical value indicating whether to return progeny results as a new assay called Progeny in the Seurat object used as input. Default to FALSE.
`...`	Additional arguments to be passed to the functions.

The publication of the method is available at: https://www.nature.com/articles/s41467-017-02391-6

The supplied expression object has to contain HGNC/MGI symbols in rows. This will, in most cases (and how we originally used it), be either normalized gene expression of a microarray experiment or log-transformed (and possible variance-stabilized) counts from an RNA-seq experiment.

The human and mouse model matrices consists of 14 pathways and large set of genes with an associated p-value (p-value per gene and pathway) that accounts for the importance of each gene on each pathway upon perturbation. Its coefficients are non-zero if the gene-pathway pair corresponds to the top N genes (100 by default) that were up-regulated upon stimulation of the pathway in a wide range of experiments. The value corresponds to the fitted z-score across experiments in our model fit. Only rows with at least one non-zero coefficient were included, as the rest is not used to infer pathway activity.

A matrix with samples in columns and pathways in rows. In case we run the method with permutations and the option get_nulldist to TRUE, we will get a list with two elements. The first element is the matrix with the pathway activity as before. The second elements is the null distributions that we generate to assess the signifance of the pathways scores.

progenyPerm

# use example gene expression matrix here, this is just for illustration
gene_expression <- as.matrix(read.csv(system.file("extdata", 
"human_input.csv", package = "progeny"), row.names = 1))

# calculate pathway activities
pathways <- progeny(gene_expression, scale=TRUE, 
    organism="Human", top = 100, perm = 1)