R-CMD-check Codecov test coverage License: GPL v3

CRAN version CRAN Downloads Hits

R package for the analysis of biological networks.

Crosstalkr provides a unified toolkit for drug target and disease subnetwork identification. Crosstalkr enables users to download and leverage high-quality protein-protein interaction networks from online repositories. Users can then filter these large networks into manageable subnetworks. Finally, users can perform in-silico repression experiments to assess the relative importance of nodes in their network.

PPI ingestion and customization

Crosstalkr allows direct access to the STRINGDB and Biogrid PPI resources. Thanks to integration with stringdb, users can evaluate biological networks from 1540 different species. For a list of supported species - just call:


Users can also take the intersection or union of the stringdb and biogrid PPIs.

Graph Filtering Methods

Crosstalkr faciliates graph reduction based on node value ranks. Node values can be provided by the user (as in gfilter.value). Users can also specify any method found in the igraph package that generates node values (i.e. igraph::degree or igraph::betweenness). We also provide a custom method for node ranking that we developed in our lab, termed network potential ( Crosstalkr provides a general implementation of a random-walk with restarts on graph structured data. We also provide user-friendly implementations of the common use-case of using random-walk with restarts to identify subnetworks of biological protein-protein interaction databases (adapted from the method described here - Given a user-defined set of seed proteins, the main compute_crosstalk function will compute affinity scores for all other proteins in the network. It will then compute a null distribution using a permutation test and compare the computed affinity scores to the null distribution to identify proteins with a statistically significant association to the user-defined seed-proteins.

Node ranking via in-silico repression

In silico repression is implemented by the node_repression function. Users must specify a state function that scores nodes. Each node in v_rm, will be systematically removed from the network. The provide state function will be applied to re-calculate network state and then the difference in total state value (sum of all nodes) will be computed.

See for more details on in-silico repression.


To install, use the following code:


For the latest development version:

install.packages("remotes") #can skip if already installed 

Given a set of user-provided set of seeds, crosstalkr will identify enriched an enriched subgraph where all nodes have a high affinity for the provided seeds.

crosstalkr is optimized for use with the human cell signaling network. For example, running the code below will return a dataframe containing the user-provided seeds as well as all other proteins in the human protein-protein interaction network with a statistically significant association to these genes.

compute_crosstalk(c("EGFR", "KRAS"))

Users can use any other kind of graph-structured data, provided they are stored in an igraph object. For example:

g <- igraph::sample_gnp(n = 1000, p = 10/1000)
compute_crosstalk(c(1,3,5,8,10), g = g, use_ppi = FALSE)

We also provide rudimentary plotting functions to allow users to quickly assess the identified subnetworks:

ct_df <- compute_crosstalk(c("EGFR", "KRAS"))

A more detailed overview of the available functionality is provided in the introductory vignette (under development).

vignette(package = "crosstalkr")

Please use the provided biorxiv pre-print to cite.


Please feel free to submit issues here. You can also contact me at if you have any questions.