Figure 3 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
Year: 2025
Type: other
Abstract: Pan-cancer drug–target interaction network. A, The network includes 854 drug–gene pairs identified by our pipeline through pairwise combinations of 214 oncogenes and 144 FDA-approved oncology drugs as high-confidence drug–gene interactions. For each identified pair, the pipeline summarizes the major cancer type(s) based on which the inference was made from the retrieved information. In the network, circles and squares represent drugs and genes, respectively. Edge colors indicate the type of interaction. Node sizes are proportional to the degree of connectivity (i.e., the number of interactions), with names labeled for genes and drugs with more than 10 connections. Drugs and genes discussed in the main text are highlighted in bold. B, Distributions of interaction types (direct binding or indirect modulation) and cancer type specificity (interactions present in a single or multiple cancer types). C, Distribution of cancer types across 25 primary sites and an additional “other” group of all other cancers. Note that each drug–gene interaction may be associated with multiple cancer types. HNC, head and neck cancer; NBL, neuroblastoma; UEC, uterine/endometrial cancer. D, Distribution of drug mechanisms of action and target specificity, which categorizes drugs as single-target (interacting with only one of the 214 genes) or multitarget (interacting with multiple genes).
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4410344011