Unlocking the Regulatory Genome by ARGUS: An Evidence-Constrained Agentic Framework for Interpreting Single Nucleotide Variants

Over 90% of disease-associated variants from genome-wide association studies fall in noncoding regulatory regions, yet their functional interpretation remains a central open problem in genomic medicine. Large language models prompted to interpret such variants routinely hallucinate transcription factor (TF) binding changes, fabricate experimental support, and assign biological significance to statistically negligible signals. We present ARGUS (Agentic Regulatory Genomics for an Uncertainty-aware Scientist), which strictly separates deterministic biological computation from LLM-mediated reasoning. ARGUS wraps 458 DNABERT-based TF binding models in a hypothesis-directed investigation loop where a planner selects evidence sources based on current uncertainty, a verifier deterministically interprets each observation, and intermediate results change the investigation path. On variant rs6983267 at the 8q24 cancer risk locus, the same planner produces four categorically different trajectories for four TFs. FOXA1 is rescued in 3 steps when real ADASTRA allele-specific binding data (15 experiments, FDR = 0.030) reveals a model false negative masked by saturation (both alleles predicted at near-maximal binding, leaving the model unable to resolve allelic differences). KLF6 traverses 8 steps across ADASTRA, JASPAR motif analysis, and ENCODE cCRE regulatory annotation before abstaining due to mixed indirect evidence. RAD21 is contradicted in 3 steps when ADASTRA finds allele-specific binding the model missed, and SP1 correctly abstains despite a log-odds ratio of 3.02 because both binding probabilities fall below the 0.5 active-binding threshold. All trajectories are produced from real data against local ADASTRA, JASPAR, and ENCODE cCRE evidence sources, with no simulated observations. A comparison of fixed-priority and LLM-mediated planning shows that the LLM planner makes better budget-allocation decisions when intermediate evidence creates asymmetries the fixed policy cannot distinguish.