Track genomics, neuroscience and molecular biology.
Search papers, labs, and topics across Lattice.
OmicsBench introduces a novel long-horizon, multi-turn benchmark for evaluating AI agents in complex bioinformatics research studies, addressing the limitations of existing single-shot task evaluations. The benchmark comprises 97 tasks with 662 dependent subtasks, designed to mimic real-world research workflows and graded deterministically, revealing significant performance gaps for current agents. An unsupervised failure mode discovery pipeline further elucidates qualitative shortcomings in agent reasoning and execution during these multi-turn analyses.
Current AI agents struggle with multi-turn bioinformatics research tasks, forgetting context and exhibiting emergent failure modes like misapplying statistical filters or hallucinating answers.
Computational biologists increasingly delegate substantial analyses to AI agents in long-horizon, multi-turn sessions, yet most benchmarks of agents'bioinformatics capabilities pose a single short task with a single-shot prompt. We present OmicsBench, a long-horizon, multi-turn bioinformatics benchmark for AI agents. Each of its 97 tasks links dependent subtasks into a directed acyclic graph, for 662 underlying subtasks in all. Each subtask is posed as one user turn and graded deterministically against an answer key, with optional in-context feedback on incorrect answers. Across 12 models on three harnesses, the slowest agent averages 3.6 hours per task, and no combination comes near saturating the benchmark, with mean scores of 0.43 to 0.57 out of a possible 1 without in-context feedback and 0.54 to 0.67 with it. Multi-turn evaluation also uncovers behavior that single-shot prompts cannot, as agents score lower on subtasks that depend on earlier subtasks, and in-context feedback helps agents recover some of that loss. Because deterministic scoring gives no qualitative account of how an agent failed, we also introduce an unsupervised failure mode discovery pipeline that describes, embeds, and clusters agents'failed attempts to surface emergent failure modes. Applied to OmicsBench, this pipeline exposes qualitative shortcomings in how agents carry out bioinformatics analyses, ranging from agents applying their own statistical filters in place of the specified ones to agents writing whole answers from memory.