Fig. 4: Network analysis supports the role of rare germline variation in somatic mutational processes.

Data in all panels were generated using physical protein interactions from the STRING database that have a combined score ≥ 80%. In panels a–d, p-values were calculated via randomization using a one-sided test. a Number of physical interactions in a random subset of the tested gene set, controlled for interaction node degree (x-axis, blue bars). Red line shows the number of interactions within genes which replicated at a 1% false discovery rate (FDR). b Number of randomly selected genes from the tested gene set interacting with at least one gene, which replicated at a 1% FDR (x-axis, red bars), controlled for interaction node degree. Red line shows the number of genes, out of the ones which additionally replicated at a 2% FDR, interacting with at least one gene replicating at a 1% FDR. c–d Same as panels a and b, after excluding known genes from the analysis (BRCA1, BRCA2, PALB2, MSH2, and MLH1). e Visualization of physical interactions between proteins corresponding to genes replicating at 1% FDR (square) and at 2% FDR (ellipse). Color code in pie chart shows the somatic mutation components the corresponding gene was associated with (bottom panel). Line width corresponds to combined (experimental, database, and text mining) STRING interaction score. Data are provided as a Source Data file.