snkit's link_nodes_to_nearest_edge is called several times and takes a while for large networks. The function doing most of the work is split_edges_at_nodes. SNKIT_PARALLEL can be set to 1 or TRUE to enable parallelism in this function, but it is not constrained (uses os.cpu_count()), so open-gira's snakemake workflow.cores limit is not respected.
Change snkit parallelism to specify an integer number of processes via SNKIT_PARALLEL or SNKIT_PROCESSES?
Use rule.threads = to specify a maximum number for any given rule invocation. Then hopefully jobs could be submitted that respect the workflow cores limit (test this...).
snkit's link_nodes_to_nearest_edge is called several times and takes a while for large networks. The function doing most of the work is split_edges_at_nodes. SNKIT_PARALLEL can be set to 1 or TRUE to enable parallelism in this function, but it is not constrained (uses os.cpu_count()), so open-gira's snakemake workflow.cores limit is not respected.