Scientific Computing and Data Management

The paper introduces a benchmark for scientific coding agents in spatial transcriptomics alignment, showing that richer context improves tool exploration but reduces alignment accuracy.

The paper re-architected a variant caller into a client-server system to improve agent-driven variant calling efficiency and reliability.

StickForStats automatically validates statistical assumptions in computational biology analyses, improving reproducibility by default-checking assumptions and rerouting to nonparametric methods when violations occur.

ScriptManager v1.0 provides a unified framework for reproducible and scalable analysis of genomics and epigenomics data across diverse assays.

FlowBench evaluates agentic bioinformatics systems by decomposing performance into planning, fault recovery, interpretation, and output fidelity, revealing limitations in toolchain inference and fault handling.

OmicOS provides an agent-operable omics ecosystem that enables AI systems to execute and verify analyses using community methods, achieving high performance in benchmark tasks.

AutoZyme optimizes bioinformatics software by autonomously identifying and applying code changes to improve runtime without significant memory use.

oxo-flow is a Rust-based workflow engine that compiles to a native binary, offering faster execution and lower memory usage compared to existing tools like Snakemake and Nextflow.

GEOAgent retrieves and preprocesses GEO datasets using AI and automation, achieving high precision and accuracy in dataset retrieval and assay classification.

GLEAM provides a web-based toolkit for accessible and rigorous machine learning analyses in biomedical research, producing accurate models across three tasks.

The KBase Research Agent automates genome analysis workflows, producing reproducible results and annotated genomes without human intervention.

The paper introduces Autopipe, a platform that enables reproducible and shareable bioinformatics pipelines generated from natural-language prompts using LLMs.

PromptBio-Bench evaluates LLM-based bioinformatics agents across 244 tasks, showing comparable performance between Biomni and ToolsGenie but declining accuracy with task difficulty.

BasalCell provides a project scaffold generator to integrate R and Python ecosystems, enabling reproducible, multi-language bioinformatics workflows with version control and CI/CD.

CARIBOU is a multi-agent AI system that autonomously performs bioinformatics analyses using HPC environments, adapting and refining workflows across datasets.

Slivka and Slivka-bio provide a lightweight framework for deploying bioinformatics tools as web services, enabling easy execution of code via HTTP and demonstrating with 15+ protein and RNA analysis tools.

BioGAIP automates complex bioinformatics tasks using an LLM-powered multi-agent system, making advanced analysis accessible to non-experts through natural language input and graphical interfaces.

This paper demonstrates how integrating AbSolution, ENCORE, and R packages can automate reproducibility in interactive applications by enabling single-step export of data, code, and documentation.

The paper evaluates open LLMs for agentic analysis in biomedical labs, finding that qwen3.6:27b matches frontier accuracy on variant calling tasks using a local setup.

ClaroAI-Bench evaluates AI agents' ability to reproduce computational findings from biomedical papers by testing data findability, accessibility, code availability, environment reconstructability, and results reproducibility.

The paper introduces Open-Rosalind, a tool-first biomedical agent system with a benchmark evaluating task accuracy, tool correctness, and trace completeness, showing tool-first execution improves performance.

This paper introduces BPL, a compiler-verified language for biology protocols, and demonstrates its ability to convert natural language SOPs into reproducible, physically valid protocols with high fidelity.

This paper introduces CompBioBench, a benchmark for evaluating agentic systems in computational biology, showing strong performance across diverse tasks.

The paper evaluates three large language model configurations on a curated bioinformatics benchmark, finding they achieve high accuracy with proper scientific skill integration and external resource access.

The paper introduces a framework for creating reusable, well-documented analysis tools in vision science, demonstrating their use in published research.

FlowWeb provides a web-based platform for flow cytometry analysis, offering automated workflows and visualization tools comparable to commercial software.

GraphMana manages population genomics data using a graph database, enabling efficient incremental updates and export to multiple formats.

RapCluster provides an interactive web platform with 11 clustering algorithms to improve transparency and reproducibility in clustering analysis.

BioClaw enables human-bot collaboration in group chats to execute biomedical analyses using containerized tools and workflows across multiple domains and data types.

Pipette is a multi-agent AI framework that generates reproducible bioinformatics workflows using a literature-derived Skill Graph, outperforming LLMs without such constraints across multiple domains.