SEMgsa {SEMgraph}

SEM-based gene set analysis

Description

Gene Set Analysis (GSA) via self-contained test for group effect on signaling (directed) pathways based on SEM. The core of the methodology is implemented in the RICF algorithm of SEMrun(), recovering from RICF output node-specific group effect p-values, and Brown’s combined permutation p-values of node activation and inhibition.

Usage

SEMgsa(g = list(), data, group, method = "BH", alpha = 0.05, n_rep = 1000, ...)

Arguments

g

A list of pathways to be tested.

data

A matrix or data.frame. Rows correspond to subjects, and columns to graph nodes (variables).

group

A binary vector. This vector must be as long as the number of subjects. Each vector element must be 1 for cases and 0 for control subjects.

method

Multiple testing correction method. One of the values available in p.adjust. By default, method is set to "BH" (i.e., Benjamini-Hochberg correction).

alpha

Gene set test significance level (default = 0.05).

n_rep

Number of randomization replicates (default = 1000).

...

Currently ignored.

Details

For gaining more biological insights into the functional roles of pre-defined subsets of genes, node perturbation obtained from RICF fitting has been combined with up- or down-regulation of genes from KEGG to obtain overall pathway perturbation as follows:

• The node perturbation is defined as activated when the minimum among the p-values is positive; if negative, the status is inhibited.

• Up- or down- regulation of genes (derived from KEGG database) has been obtained from the weighted adjacency matrix of each pathway as column sum of weights over each source node. If the overall sum of node weights is below 1, the pathway is flagged as down-regulated otherwise as up-regulated.

• The combination between these two quantities allows to define the direction (up or down) of gene perturbation. Up- or down regulated gene status, associated with node inhibition, indicates a decrease in activation (or increase in inhibition) in cases with respect to control group. Conversely, up- or down regulated gene status, associated with node activation, indicates an increase in activation (or decrease in inhibition) in cases with respect to control group.

Value

A list of 2 objects:

1. "gsa", A data.frame reporting the following information for each pathway in the input list:

   • "No.nodes", pathway size (number of nodes);

   • "No.DEGs", number of differential espression genes (DEGs) within the pathway, after multiple test correction with one of the methods available in p.adjust;

   • "pert", pathway perturbation status (see details);

   • "pNA", Brown's combined P-value of pathway node activation;

   • "pNI", Brown's combined P-value of pathway node inhibition;

   • "PVAL", Bonferroni combined P-value of pNA, and pNI; i.e., 2* min(pNA, PNI);

   • "ADJP", Adjusted Bonferroni P-value of pathway perturbation; i.e., min(No.pathways * PVAL; 1).

2. "DEG", a list with DEGs names per pathways.

Author(s)

Mario Grassi [e-mail]

References

Grassi, M., Tarantino, B. SEMgsa: topology-based pathway enrichment analysis with structural equation models. BMC Bioinformatics 23, 344 (2022). https://doi.org/10.1186/s12859-022-04884-8

Examples

## Not run: 

# Nonparanormal(npn) transformation
als.npn <- transformData(alsData$exprs)$data

# Selection of FTD-ALS pathways from kegg.pathways.Rdata

paths.name <- c("MAPK signaling pathway",
                "Protein processing in endoplasmic reticulum",
                "Endocytosis",
                "Wnt signaling pathway",
                "Neurotrophin signaling pathway",
                "Amyotrophic lateral sclerosis")

j <- which(names(kegg.pathways) %in% paths.name)

GSA <- SEMgsa(kegg.pathways[j], als.npn, alsData$group,
              method = "bonferroni", alpha = 0.05,
              n_rep = 1000)
GSA$gsa
GSA$DEG


## End(Not run)