What is pathfindR?
pathfindR is an R package for enrichment analysis via active
subnetworks. Conventional enrichment methods ignore protein–protein
interaction (PPI) information, which can leave the picture incomplete.
pathfindR’s main function identifies active subnetworks in a
protein-protein interaction network using a user-provided list of genes
and associated p values, then performs enrichment analyses on those
subnetworks to identify enriched terms (pathways, or more broadly, gene
sets) that may underlie the phenotype of interest.
Beyond the core workflow, pathfindR also lets you:
- cluster the enriched terms and identify a representative term in each cluster,
- score the enriched terms on a per-sample basis, and
- compare two pathfindR results against each other.
The methods are described in detail in Ulgen E, Ozisik O, Sezerman OU (2019), pathfindR: An R Package for Comprehensive Identification of Enriched Pathways in Omics Data Through Active Subnetworks, Front. Genet. (doi:10.3389/fgene.2019.00858).
What’s new?
The headline of pathfindR 3.0.0 is something you mostly won’t notice: the Java dependency is gone. Active subnetwork search - the step that grows candidate subnetworks over a protein–protein interaction network - has been re-implemented entirely in R and C++ (via Rcpp), replacing the old Java JAR that pathfindR used to shell out to.
In practice this means no JDK, and no “Java not found” issues on a fresh machine. You install the package and it runs:
install.packages("pathfindR") # or remotes::install_github("egeulgen/pathfindR")
library(pathfindR)
output <- run_pathfindR(example_pathfindR_input)
The greedy and simulated-annealing search methods produce results that are numerically identical to the old JAR given the same seed and inputs, so existing analyses remain reproducible - you’re just getting them without the extra runtime to install and configure.
A few renames
Some exported functions were renamed to follow a more consistent, snake_case style. If you call these directly in scripts, update the names:
-
active_snw_search()→get_active_subnetworks()(its signature changed too) -
filterActiveSnws()→filter_active_subnetworks()
The most common entry point, run_pathfindR(), is unchanged.
Term–gene graph, split in two
The old term_gene_graph() did two jobs at once: build the graph and
draw it. Those are now separate, which makes it much easier to tweak the
visualization or reuse the underlying structure:
create_term_gene_graph()builds theigraphobject.create_term_gene_plot()renders a plot from that object.
g <- create_term_gene_graph(output, num_terms = 10)
create_term_gene_plot(g)
Should you upgrade?
Yes! especially if Java setup has ever bitten you or your collaborators. The science is the same, the results are the same, and the install is now a single, self-contained step. Full notes are in the changelog.