Implementation of hypergraphlet counter used in
Gaudelet, T., Malod-Dognin, N. and Pržulj, N., 2018. Higher-order molecular organization as a source of biological function. Bioinformatics, 34(17), pp.i944-i953.
It is now updated to output both the hypergraphlets and the simplets counts. Simplets were introduced in
Malod-Dognin, N. and Pržulj, N., 2019. Functional geometry of protein interactomes. Bioinformatics, 35(19), pp.3727-3734.
Both hypergraphlets and simplets give features to characterize the wiring patterns around nodes in hypergraphs (or simplicial complexes). The simplets are a restriction of hypergraphlets and give a smaller signature. We refer the reader to both papers above for more information.
The implementation requires BOOST library (https://www.boost.org/) and a compiler supporting C++11.
Before compiling add the absolute path to the BOOST lib/ folder on your system to the Makefile
Compile by opening a terminal, navigating to the hypergraphlets folder, and entering the command make.
To run the counter simply enter the command
./run_hypercounter -g "path_to_file/hyperedge.list" -o "path_to_output/name_of_output" -t "Number of threads, default 1" -b "Number of groups of genes to launch across the threads, default 50"
The hyperedge list should follow the format: - tab separated - each row starts with the hyperedge index (between 0 and m-1, where m is the total number of hyperedges) and contain the list of vertices it contains - vertices should be indexed from 0 to n-1, n being the total number of vertices
The counter outputs two .svml files containing the hypergraphlets and simplets counts with the following format: - each row correspond to the count for a vertex - the first number (and last) correspond to the vertex index - the count for each orbit is of the shape orbit:count
The code uses OpenMP for parallelisation, the number of threads to use can be set using the flag -t.
The vertex sets is partitioned in a number of blocks (set by the flag -b) each block being processed by a single thread. The partition is randomised to avoid having consecutive vertex in the same block (this is to avoid having the densely connected vertices in the same block)