Identifying Multi-Hit Cancer Drivers Without Massive Parallelization: A CP, MIP, and Column Generation Framework

We introduce RegNetAgents, an AI-oriented multi-agent framework for structured, query-driven regulatory candidate identification across heterogeneous gene regulatory networks. It integrates bulk tumor (TCGA) and single-cell (GREmLN project) ARACNe networks and labels each candidate regulator by the network or networks in which it appears (Both, TCGA-only, GREmLN-only). For a given focal gene, the framework finds its regulators in both networks and labels each by source, flags those that are known cancer driver genes (IntOGen), and, for tumor-network regulators, gives the mode of action (MoA; activating or repressive). It is implemented as a multi-agent LangGraph state-graph workflow, accessible through a Python API and a Model Context Protocol (MCP) client, and operates as a downstream analytical layer over precomputed networks rather than a network inference method. For example, a single query for CTNNB1 in BRCA returns two tumor-specific (TCGA-only) driver-gene regulators, DDR2 and IL6ST, both with activating MoA. Across twelve breast cancer (BRCA) and thirteen colorectal cancer (COAD) driver genes, we compared each gene's tumor-network-only (TCGA-only) regulators with the TCGA-only regulators of random non-driver genes. Regulators as a group are about 2.7-fold richer in driver genes than genes overall, so almost any gene's regulators look enriched when tested against all genes. We therefore used random genes as the baseline. In COAD, the cancer genes' regulators include modestly but consistently more IntOGen driver genes than random genes' regulators do (nominal p = 0.012-0.036 across five random samples); in BRCA the difference is borderline (p = 0.048-0.083). Housekeeping and non-driver control genes show no such excess in the tumor-network tier, and the same comparison for GREmLN-only regulators shows none (p >= 0.20). Code: https://github.com/jab57/RegNetAgents

In this paper, we formulate the new multi-objective coverage (MOC) problem where our goal is to identify a small set of representative samples whose predicted outcomes broadly cover the feasible multi-objective space. This problem is of great importance in many critical real-world applications, e.g., drug discovery and materials design, as this representative set can be evaluated much faster than the whole feasible set, thus significantly accelerating the scientific discovery process. Existing works cannot be directly applied as they either focus on sample space coverage or multi-objective optimization that targets the Pareto front. However, chemically diverse samples often yield identical objective profiles, and safety constraints are usually defined on the objectives. To solve this MOC problem, we propose a novel search algorithm, MOC-CAS, which employs an upper confidence bound-based acquisition function to select optimistic samples guided by Gaussian process posterior predictions. For enabling efficient optimization, we develop a smoothed relaxation of the hard feasibility test and derive an approximate optimizer. Compared to the competitive baselines, we show that our MOC-CAS empirically achieves superior performances across large-scale protein-target datasets for SARS-CoV-2 and cancer, each assessed on five objectives derived from SMILES-based features.

Pareto optimization via evolutionary multi-objective algorithms has been shown to efficiently solve constrained monotone submodular functions. Traditionally when solving multiple problems, the algorithm is run for each problem separately. We introduce multitasking formulations of these problems that are an effective way to solve multiple related problems with a single run. In our setting the given problems share a monotone submodular function $f$ but have different knapsack constraints. We examine the case where elements within a constraint have the same cost and show that our multitasking formulations result in small Pareto fronts. This allows the population to share solutions between all problems leading to significant improvements compared to running several classical approaches independently. Using rigorous runtime analysis, we analyze the expected time until the introduced multitasking approaches obtain a $(1-1/e)$-approximation for each of the given problems. Our experimental investigations for the maximum coverage problem give further insight into the dynamics behind how the approach works and doesn't work in practice for problems where elements within a constraint also have varied costs.