AIPOCH Open-Science v0.31.1 发布：新增 ENA 测序数据、g:Profiler 富集与 NCBI 基因组查询

Official release notes (https://github.com/aipoch/open-science/releases/tag/v0.31.1) # AIPOCH Open-Science v0.31.1 > A reach-and-reliability release: query ENA sequencing runs and archive-generated FASTQ files by accession, run g:Profiler gene-set enrichment and NCBI reference-genome lookups, connect an optional dedicated classification model, and pair remote access more safely — with a long tail of notebook, session, and connector fixes. AIPOCH Open-Science is an open-source, local-first, model-agnostic AI research workbench for scientists and researchers. It enables reproducible, inspectable research across models with scientific AI agents, Python and R execution, scientific data connectors, and cross-platform support for macOS, Windows, and Linux. v0.31.1 widens what a session can reach and steadies everyday work. New ENA tools resolve a public ENA/INSDC study, experiment, sample, or run accession to its sequencing runs, returning study, sample, experiment, organism, platform, and library metadata together with the archive-generated FASTQ files for downstream analysis. The Genes connector gains g:Profiler-backed GO and pathway enrichment for differential or marker gene sets, with organism scoping, selectable evidence sources, and an explicit statistical background. Standalone NCBI tools resolve taxon names with ambiguity reporting, inspect versioned genome assemblies, and look up sequence aliases, while historical assembly accessions stay queryable instead of being replaced by the latest revision. Settings can optionally connect a dedicated classification model service, so skill and connector capability selection runs on a purpose-chosen model instead of a conversation model — the feature stays off until a service is added, and saved credentials are verified before they stick. Remote-access pairing requests appear ahead of the trusted-browser list with countdowns and urgent badges, and revoking trusted browsers is safe even when the current browser revokes itself. An R run blocked by notebook network protection now shows an inline warning card that links to the setting that resolves it. And the long tail is real: persistent Windows kernels are supervised so runs end cleanly, stale package inventories retry, lost OpenCode sessions recover on startup, duplicate resume ownership is guarded, approval cards stop reopening and stealing composer focus, GTEx results paginate instead of truncating, Open Targets partial errors surface, literature figures and table structure recover from PDFs, and hg19 conservation analysis defaults to the correct track. ## ✨ Highlights - **Reach sequencing runs by accession.** New ENA tools resolve a public ENA/INSDC study, experiment, sample, or run accession to its sequencing runs — returning study, sample, experiment, organism, platform, and library metadata together with the archive-generated FASTQ files ready for downstream analysis. (#2813) - **Gene-set enrichment powered by g:Profiler.** The Genes connector can now run GO and pathway enrichment for a differential or marker gene set, with organism scoping, selectable evidence sources, and an explicit statistical background so enrichment results are well controlled. (#2802) - **NCBI reference genome lookups.** Standalone NCBI tools resolve taxon names with ambiguity reporting, inspect versioned genome assemblies, and look up sequence aliases — historical assembly accessions stay queryable instead of being replaced by the latest revision. (#2796) - **Optional classification models.** Settings can connect a dedicated classification service so skill and connector capability selection runs on a purpose-chosen model instead of a conversation model; the feature stays off until a service is added, and saved credentials are verified before they stick. (#2799) ## 🚀 New Features - ENA run and FASTQ lookup: an ENA/INSDC accession in any supported form (study, experiment, sample, or run) resolves to its sequencing runs with study, sample, experiment, organism, platform, and library metadata, plus the archive-generated FASTQ files for downstream analysis. (#2813) - g:Profiler gene-set enrichment: discover supported organisms, namespaces, and source data versions, then run GO and pathway enrichment for a gene set with an explicit organism, optional source selection, custom statistical background, and adjustable significance controls. (#2802) - NCBI reference genome tools: paginated taxon resolution with ambiguity reporting, versioned assembly inspection that keeps historical accessions queryable, and sequence-alias lookup ahead of downstream analysis. (#2796) - Optional classification models: a new Settings panel connects a dedicated classification service; a single automatic-capability binding covers skills and connectors, models are selected from the adapter catalog, and credentials are probed before saving. (#2799) - Remote-access pairing requests appear ahead of the trusted-browser list, with countdowns, urgent badges, and matching-code guidance — and revoking trusted browsers is safe even when the current browser revokes itself. (#2812) - An R run blocked by notebook network protection shows an inline warning card that links to the network setting that resolves it and explains that the cell was not executed. (#2790) ## 🔧 Improvements - Long conversations stay responsive: transcript history layout and streaming message parsing are bounded, and annotation and resize subscription work is stabilized. (#2801, #2807) - Notebook run history traces REPL dependencies and records the file lineage produced by a session handoff. (#2808) - Classification runs record per-call diagnostics so silent model failures become visible. (#2806) ## 🐛 Bug Fixes - **Notebook and compute** — persistent Windows kernel process trees are supervised so runs end cleanly (#2770); stale Windows R package inventories retry instead of failing (#2789); package inventory reports from micromamba environments are accepted (#2774); REPL termination records diagnostics (#2777); shell commands can access GUI-authorized external folders (#2797); packaged notebook scripts stop picking up unrelated module definitions (#2781). - **Sessions and agent runtime** — OpenCode sessions that went missing are recovered on startup (#2805); duplicate session-resume ownership is guarded (#2810); cancellation is preserved and unreaped delegates are quarantined (#2518); sessions with missing Codex rollout files are adopted (#2779); reliable parent message continuations are admitted (#2803); delegate ownership records recover after an app restart (#2785); session identity is preserved across framework transitions (#2778). - **Connectors** — GTEx gene reference results paginate instead of truncating (#2768); Open Targets GraphQL partial errors are surfaced instead of silently dropped (#2760); hg19 conservation analysis selects the correct default track (#2796). - **Interface and storage** — approval cards no longer reopen or steal composer focus (#2654); duplicate pending message previews are removed (#2655); the recent-session list keeps its order while sessions update (#2769); keyboard access regressions are addressed (#2786); passive runtime observers no longer conflict with session storage (#2649); OpenCode publication recovers when a file permission error interrupts it (#2780); skill-creator helpers stop being confused with unrelated module definitions (#2804); literature figures and table structure recover correctly from PDFs (#2794). ## 📦 Install **Requirements:** macOS 12+ (Apple Silicon or Intel), Linux x64, or Windows 10/11 x64. On first run, the onboarding wizard checks the environment and can install and configure an app-managed agent runtime. Once installed, the app can update itself in place. Download the appropriate package from the **Assets** section below: | Platform | Package | | --------------------- | ---------------------------------- | | macOS (Apple Silicon) | DMG for ARM64 | | macOS (Intel) | DMG for x64 | | Linux | AppImage or Debian package for x64 | | Windows | Installer for x64 | **macOS — first launch.** Official release builds are Developer ID signed and notarized by Apple, so they open like other trusted applications. A locally built copy is not notarized and may require approval through macOS Privacy & Security. **Windows — first launch (unsigned build).** No Authenticode certificate yet, so SmartScreen shows a bypassable "unrecognized app" prompt (**More info → Run anyway**). Verify that the package came from the official release page before continuing. **Build from source instead:** npm install npm run build:mac # or: build:linux / build:win ## 🧭 What's in this release (maturity) - ✅ **Implemented:** everything shipping in v0.31.0, plus: ENA run resolution with archive-generated FASTQ files; g:Profiler GO and pathway enrichment with custom statistical backgrounds; NCBI taxon, assembly, and sequence-alias lookups that keep historical accessions queryable; an optional dedicated classification service with automatic capability selection; prioritized remote-access pairing with safe trusted-browser revocation; inline guidance when network protection blocks an R run; and bounded transcript layout and streaming parse work for long conversations. - 🚧 **Partial:** provider choice remains constrained by the active framework's endpoint compatibility; remote compute covers direct SSH and Slurm (cloud-GPU submission is not built yet); WSL2 Bash remains an opt-in Windows x64 preview; the skills commons ships marketplace discovery but not cross-machine forking or user-facing version pinning; reproducibility ships replayable per-version verification, while full-session replay and solver-exact equivalence remain open; conditional package restore is an exported script workflow; and review is opt-in and record-scoped. - 🗺️ **Roadmap:** a unified model gateway, cross-machine skill forking and version pinning, cloud-GPU execution, stronger sandboxing and credential isolation, and collaborative research workflows. ## 🐢 Known Limitations - **Verification checks a captured version, not a whole session.** Full-fidelity session replay and solver-exact equivalence guarantees remain open, and starting a check from the web client currently requires the desktop app. - **The Windows reset utility is deliberately conservative.** It refuses protected roots, network paths, nonstandard custom roots, and reparse points; uninspectable or running runtimes block cleanup. - **Conditional package restore is an exported workflow.** The restore script recreates packages into a destination you own; the app does not adopt or manage restored libraries. - **PDF extraction depends on locally installed model resources.** First-run analysis downloads verified resources; cached results reopen without re-analysis. - **Imported `.science` sessions are read-only.** They stay inspectable and referenceable, and forking them into a writable continuation remains available; execution and continuation remain disabled on the import itself. - **Bookmarks belong to a single session.** They do not follow branch switches, join `.science` package exports, or sync across machines — forking copies them into the new session with fresh identities. - **WSL2 Bash is a preview.** It is opt-in on Windows x64 only; PowerShell remains the default shell runtime. - **Remote compute covers direct SSH and Slurm.** Cloud-GPU submission is not built yet. - **Editable artifacts cover text formats.** Binary formats stay read-only, and editing always publishes new versions. - **Network sandboxing covers the app's notebook and compute runtimes, not the whole system.** On Windows, the boundary applies only after the sandbox's one-time administrator setup. - **Provider choice is per framework, not one unified gateway.** The available protocol depends on the selected agent backend, and remote endpoints must use HTTPS. - **Code reconstruction is LLM-generated.** It does not replace deterministic reproduction; replayable verification is the deterministic check. - **The reviewer is opt-in and record-scoped.** It does not replace domain-specific validation of citations, units, statistics, or methods. - **Windows builds are unsigned.** SmartScreen may warn on first launch; official macOS builds are notarized. - **No local GPU compute backend.** - **No multi-user real-time collaboration.** ## 🙏 Acknowledgements Thanks to @ewen-poch, @wen2zhou, @nasus2002, @Ada.Liu, and everyone in [Discord](https://discord.gg/zxQAYjReRv), [X](https://x.com/aipoch_ai), and [Discussions](https://github.com/aipoch/open-science/discussions). --- **Full Changelog**: https://github.com/aipoch/open-science/commits/v0.31.1 Repository README (current repository documentation, not release changes) AIPOCH Open-Science AI research workbench for reproducible science — open-source, local-first, and model-agnostic. AIPOCH Open-Science is an AI research workbench for scientists and researchers, developed by [AIPOCH](https://aipoch.com/open-science) with an open-source, local-first, model-agnostic approach. It enables reproducible, inspectable research with scientific AI agents, Python and R execution, scientific data connectors, and cross-platform support for macOS, Windows, and Linux. Create a project, describe your research goal in plain language, and let the agents read files, search the web, run code, query scientific data sources, and produce reports, tables, and figures with traceable provenance—all in one workspace. AIPOCH Open-Science supports computational and data-intensive research across disciplines, including machine learning, statistics, life sciences, chemistry, materials science, physics and environmental science. It supports the research process from literature review and hypothesis development to code execution, data analysis, simulation, visualization, and the production of traceable research outputs. > 💡 **[AIPOCH Open-Science v0.31.1 released](https://github.com/aipoch/open-science/releases/latest)** _(last updated September 2026)_. AIPOCH Open-Science v0.31.1 broadens the scientific data you can reach from a session: ENA tools resolve a public ENA/INSDC accession to its sequencing runs with archive-generated FASTQ files, the Genes connector gains g:Profiler-powered GO and pathway enrichment with a custom statistical background, and new NCBI tools resolve taxon names, inspect versioned genome assemblies, and look up sequence aliases. Settings can optionally connect a dedicated classification model service, and remote-access pairing moves to the top of the security panel with safe revocation of trusted browsers. Stability is tighter throughout — Windows R notebooks supervise persistent kernels and guide network recovery, lost OpenCode sessions recover on startup, and approval cards, pending message previews, and the recent-session list behave more reliably. See the [latest release notes](https://github.com/aipoch/open-science/releases/latest) for full details. ## Table of Contents - [Quick Start](#-quick-start) - [Product Tour](#product-tour) - [Benchmark Performance](#benchmark-performance) - [Core Capabilities](#core-capabilities) - [Model Providers](#model-providers) - [Data, Permissions, and Trust](#data-permissions-and-trust) - [Development & Packaging](#development--packaging) - [Frequently Asked Questions](#frequently-asked-questions) - [Get Involved](#get-involved) - [License](#license) - [Star History](#star-history) ## 🚀 Quick Start ### 1. Download the app Open the [latest release](https://github.com/aipoch/open-science/releases/latest), expand **Assets**, and choose the installer for your computer: | Your computer | Choose | | --------------------------------------- | ---------------------------------------- | | macOS 12+ — Apple Silicon (M1 or newer) | The macOS DMG for Apple Silicon / ARM64 | | macOS 12+ — Intel | The macOS DMG for Intel / x64 | | Windows x64 | The Windows x64 installer | | Linux x64 | The Linux x64 AppImage or Debian package | Download from the official release page; see [download verification](SECURITY.md#verifying-your-download) if needed. On macOS, you can also install with [Homebrew](https://brew.sh): ```bash brew install --cask open-science ``` Windows reinstalls preserve research data. For a full cleanup, see the [data reset tool](scripts/windows-reset/README.md), which permanently deletes local data after confirmation. ### 2. Complete first-time setup Follow the setup wizard: **Environment → Data location → Agent runtime → Model provider → Notebook runtime**. Complete the required environment and agent-runtime checks and test your model connection. Python/R Notebook setup is optional; Notebook and data-location settings can be changed later. Host compatibility, storage, and network checks Provider, API Key, endpoint, and model validation ### 3. Start a research project 1. Click **New project**, open a session, and describe your research goal, inputs, and expected outputs. 2. Attach files, select a model and approval mode, then send the task. Use `@` to reference project files or `/` to choose a skill. 3. Review tool activity and any approval requests, preview the results, and check their available evidence in **Provenance**. > Screenshots in this README illustrate the workflow. Labels, catalogs, and other interface details may differ from the version you install. ## Product Tour ### From a research request to a traceable result Consider a representative bioinformatics task: reproduce a published differential-expression analysis, compare the regenerated results with the paper, and deliver the report, tables, and figures needed for review. The screenshots below are representative views from documented AIPOCH Open-Science workflows; they illustrate each stage rather than one continuous session. #### 1. Define the research task and evidence Describe the research question, source paper and datasets, required methods or thresholds, expected outputs, and acceptance criteria. Upload supporting files or reference an existing project artifact with `@`, so the agent starts from explicit inputs instead of hidden context. #### 2. Execute with inspectable scientific tools The agent can combine scientific skills, permissioned research connectors, searches, file operations, and Python or R code in the shared Notebook. Generated figures can be reviewed beside the research summary, while the artifact record exposes captured producer code and execution evidence for inspection. #### 3. Review reports, tables, and figures in place The final response summarizes what reproduced, what differed, and which limitations matter. Generated Markdown reports, CSV tables, images, and other research artifacts remain attached to the session and are collected in the project file library, where they can be previewed beside the conversation and reused in follow-up work. #### 4. Trace every artifact back to its evidence Each generated artifact is stored as an immutable, checksummed version. Its **Provenance** view can expose the producing code and execution history, referenced inputs, observed environment inventory, producing conversation branch, and version-scoped Reviewer findings. Evidence that could not be verified is marked unavailable rather than inferred. ## Benchmark Performance ### 🏆 #1 on BiomniBench-DA Public 50 AIPOCH Open-Science achieved the highest ranking score in the compiled BiomniBench-DA Public 50 comparison, earning **79.05** with **gpt-5.6-sol (xhigh)**. The result combines a Gemini 3.1 Pro judge score of **81.04** and a DeepSeek v4-pro judge score of **77.06** through an equal-weight mean, placing AIPOCH Open-Science **#1** among the collected Public 50 results. Explore the [BiomniBench-DA dataset](https://huggingface.co/datasets/phylobio/BiomniBench-DA). ## Core Capabilities AIPOCH Open-Science combines project management, multi-model agent execution, Python and R notebooks, scientific data connectors, immutable artifact versions with provenance, and permissioned human-in-the-loop control in one local workspace. The installed app and [latest release notes](https://github.com/aipoch/open-science/releases/latest) are the source of truth for changing catalogs, packaging details, and newly added options. | Area | Core capability | | --------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Scientific skills** | Extend research workflows with **23 built-in skills** and **525 skills** available from the [Skills Marketplace](https://github.com/aipoch/openscience-skill-marketplace), with one-click installation and updates. Create skills through conversation or completed work, and import packages or GitHub sources. Marketplace contributions are published after review; local imports do not publish skills. | | **Connectors** | Access scientific resources through **24 built-in connectors**, or add custom local and remote MCP connectors. Manage tool-level permissions and import or export connector configurations. | | **Specialists and delegation** | Install **10 Specialists** from the [Specialist Marketplace](https://github.com/aipoch/openscience-specialist-marketplace), or create and customize personal specialists for delegation from the main agent. Specialist packages support import and export; marketplace contributions are reviewed before publication, and local imports do not publish them. | | **Models and agent backends** | Use cloud models, compatible custom gateways, or Claude and Codex subscription logins. Choose Claude Code, OpenCode, Codex, or CodeBuddy as the agent backend, with model connection checks, image input, and reasoning controls. | | **Projects, sessions, and research packages** | Organize projects with pinned sessions, message branches, side chats, and recoverable history. Export a **portable `.science` research package** and import it into another project or computer with conversation branches, selected file versions, Notebook records, and verification evidence. Imports are read-only and do not execute code or restore credentials; side chats and bookmarks are excluded, and included files depend on the export selection. | | **Reviewer** | Enable optional auto-review to check a completed agent turn's responses, execution logs, and related file evidence in a separate context. Get evidence-backed pass, warning, or failure checks, with a bounded cycle of main-agent corrections and re-review when issues are found. Review logs and issue-resolution states remain available; the review is limited to records available for that turn. | | **Python, R, notebooks, and HPC** | Run Python, R, Notebook, and shell workloads locally using managed environments or your own interpreters, with background execution and recorded history. Connect to remote hosts through SSH or submit work through Slurm; remote workloads require the host, software, resources, and permissions described in the remote-compute FAQ. | | **Literature Library** | Import and manage references and PDFs with collections, tags, project links, notes, and duplicate merging. Find open-access full text, read PDFs and extract figures and tables, and use library sources in conversations for AI-assisted analysis. Generate bibliographies in your chosen citation style and export references as BibTeX or RIS. | | **Scientific files and previews** | Upload individual files up to **10 GiB**, organize project files, and preview scientific data, PDFs, Office documents, images, code, and molecular structures. This upload limit does not guarantee that a model can read an entire file: model context, attachment parsing, and previews have separate limits. Large files usually need chunked reading or analysis with code. | | **Artifacts and provenance** | Keep immutable artifact versions with available producer code, inputs, execution history, environment information, and review evidence. In the desktop app, replay eligible versions with a complete recipe, required inputs, and a usable runtime, then compare outputs and export verification records. Missing evidence may prevent verification, and replay checks do not establish scientific validity. | ## Model Providers AIPOCH Open-Science is model-agnostic at the product level: connect it to major cloud LLM providers, a custom gateway, or reuse an existing Claude or Codex subscription. Provider availability currently depends on the selected agent backend and the API protocols it supports. There are four ways to connect a model: | Provider mode | How it works | | ---------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Built-in cloud providers** | Choose from the provider list shown by the installed app and authenticate with the requested key. | | **Custom Gateway** | Supply a Base URL and exact model ID with an API protocol supported by the selected agent backend (Messages, Chat Completions, or Responses), then run the connection test. Remote gateways require HTTPS and an API Key. Loopback endpoints such as `localhost`, `127.0.0.1`, or `[::1]` may use HTTP without a key; presets include Ollama, LM Studio, llama.cpp, and vLLM. A default API format does not guarantee server or model compatibility. | | **Codex Subscription** | Select the Codex agent framework, then choose Codex Subscription as the provider type. | | **Claude Subscription** | Sign in with a Claude subscription in two modes: **shared** (a browser login that stores credentials in your default `~/.claude` profile) or **isolated** (an app-managed `claude setup-token` run under an app-owned `CLAUDE_CONFIG_DIR`, fully isolated from `~/.claude/`, with a browser flow plus a paste-a-token fallback). | Built-in providers include OpenAI, Anthropic, DeepSeek, NVIDIA Build, and others; available models and regional endpoints depend on the installed version and selected agent backend. Check the provider selector and connection test in the app. ## Data, Permissions, and Trust AIPOCH Open-Science stores project data, settings, artifact versions, and provenance evidence on the local computer. API Keys are kept locally and use the operating system's secure credential storage when it is available. Logs are local and are not uploaded automatically. External data flow is still possible and should be reviewed: - Model requests send the prompt and necessary context to the selected model provider. - Web searches and remote connectors send their displayed parameters to external services. - Local connectors may execute trusted commands on the computer. - The application can also contact update servers, marketplace catalogs, and runtime or model download services. - Attachments, `@` references, logs, and generated reports may contain sensitive research data. Choose the narrowest permission profile that fits the task: | Mode | Behavior | Recommended use | | -------------------- | -------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------- | | `Ask for approval` | Requests approval for actions not already covered by scoped grants or trusted application tool policies | New workflows, sensitive data, unfamiliar scripts | | `Auto-approve edits` | Uses the backend's native auto review when available; otherwise automatically allows clearly low-risk workspace operations | Trusted file-editing work with controlled external access | | `Full access` | Automatically allows edits, commands, network, and connectors | Clearly scoped, fully trusted, unattended work | The effective profile depends on the selected backend and existing grants. Connector, tool, and compute-network policies also apply; check the effective mode shown by the app. Review connector parameters and tool activity before approving them. Never include API Keys, access tokens, patient identifiers, unpublished data, or sensitive local paths in screenshots or public issue logs. ## Development & Packaging AIPOCH Open-Science is an Electron application built with React, TypeScript, Prisma/SQLite, and an ACP-based agent runtime. Prerequisites for source development