Abstract 4139: Resource allocation deconvolution for pathway analysis

Abstract Background: Traditional pathway analysis mainly focuses on comparing pathway activity among different samples, ignoring how pathways share and compete for a limited transcriptional "budget." Methods: We present PathwaySpectra, a framework that characterizes the transcriptional budget allocation and competitive landscape of pathways at the single-sample level. It can flexibly integrate standard annotations and user-defined gene sets and can be easily extended to various data types and disease scenarios. Results: In an immune checkpoint inhibitor (ICI) cohort, PathwaySpectra revealed previously unannotated high-efficiency modules whose budget shares positively tracked clinical response, as well as low-efficiency modules enriched in progressive disease. Using composition-aware models with covariate adjustment, these associations remained robust. Compared to common supervised pathway scores, PathwaySpectra not only demonstrates the ability to distinguish response-related signals but also to identify previously unknown related signals. The competition view further suggested a reallocation of budget from immune-effective to inefficient pathways in non-responders, potentially constraining resources for productive immune activation. Conclusions: PathwaySpectra offers a complementary perspective beyond standard pathway activity metrics, enabling budget-based, sample-by-sample pathway analysis and supporting custom gene sets for research targeting specific questions, thereby facilitating the discovery of pathway-level therapeutic response biomarkers. Citation Format: Junhao Wang, Yifei Wang, Tianshu Michael Bao, Yong Li, . Resource allocation deconvolution for pathway analysis abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4139.