NEW Native Desktop App (.dmg / .deb / .rpm / .AppImage), VS Code Extension (Marketplace + Open VSX), Disk-Pressure Guard & Scientist Team AI Profile → 全新发布:原生桌面端应用、VS Code 官方插件、磁盘压力自动守护、数据树防篡改与 Scientist Team 多智能体协同 →
Compiled in Rust · #![forbid(unsafe_code)] · Single Binary · CLI + Web Studio + Desktop + VS Code IDE Rust 原生编译 · #![forbid(unsafe_code)] · 单二进制 CLI + Web 工作台 · 另有原生桌面端与 VS Code 插件

Build, verify, and scale reproducible bioinformatics workflows. 构建、验证并规模化运行可复现的生物信息学工作流。

oxo-flow compiles declarative .oxoflow TOML pipelines into parallel Directed Acyclic Graphs with per-rule isolation across 8 environment backends, a unified AI Companion & Scientist Team, runtime disk-pressure protection, and cryptographic checkpoint provenance — zero Python or JVM overhead. oxo-flow 将声明式 .oxoflow TOML 流程编译为高并发有向无环图(DAG),原生提供 8 种环境隔离后端、统一的 AI 伴侣与科学家团队多智能体、运行时磁盘压力自动回收守护以及密码学级断点溯源校验 —— 零 Python 或 JVM 启动开销。

$cargo install oxo-flow-cli
✓ validate · 0 warnings
📄 somatic-wes.oxoflow Click any DAG node → inspect rule
[workflow] name = "somatic-tumor-normal" version = "1.2.0" [engine] min_free_disk = "50G" # Auto reclaim/hold before disk full reclaim_free_disk = "120G" protected_data_trees = ["raw/", "/ref/GRCh38"]
[[rules]] name = "fastp_qc" input = ["raw/{sample}_R1.fq.gz", "raw/{sample}_R2.fq.gz"] output = ["clean/{sample}_R1.fq.gz", "clean/{sample}_R2.fq.gz", "qc/{sample}.fastp.json"] temporary = true # Eligible for Disk-Pressure cascade reclaim environment = { pixi = "envs/qc.toml" } shell = "fastp -w {threads} -i {input[0]} -I {input[1]} -o {output[0]} -O {output[1]} -j {output[2]}"
[[rules]] name = "bwa_mem2_align" input = ["clean/{sample}_R1.fq.gz", "clean/{sample}_R2.fq.gz"] output = ["bam/{sample}.sorted.bam"] threads = 16 memory = "32G" environment = { docker = "quay.io/biocontainers/bwa-mem2:2.2.1" } shell = "bwa-mem2 mem -t {threads} /ref/GRCh38.fa {input[0]} {input[1]} | samtools sort -o {output[0]}"
[[rules]] name = "qc_parse" # Domain parser auto-aggregates metrics input = ["qc/{sample}.fastp.json"] output = ["qc/{sample}.metrics.tsv"] environment = { system = true } shell = "oxo-flow report parse-qc fastp {input[0]} -o {output[0]}"
[[rules]] name = "mutect2_somatic" # Scales across [[pairs]] tumor/normal input = ["bam/{experiment}.sorted.bam", "bam/{control}.sorted.bam"] output = ["vcf/{pair_id}.somatic.vcf.gz"] environment = { singularity = "images/gatk-4.6.sif" } shell = "gatk Mutect2 -R /ref/GRCh38.fa -I {input[0]} -I {input[1]} -O {output[0]}"
[[pairs]] pair_id = "CASE_001" experiment = "EXP_01" control = "CTRL_01"
🚇 Interactive Metro DAG Graph (Click a node) petgraph · topological levels: 3
fastp_qc pixi · temp=true bwa_mem2 docker · 16c/32G qc_parse native fastp QC mutect2 [[pairs]] · sif
Rule: fastp_qc ({sample} × 24) sha256 verified

Expands wildcard {sample} across discovered paired-end FASTQ inputs. Marked temporary = true so the Disk-Pressure Guard can tombstone intermediate trimmed reads once downstream BAMs complete.

Backend: pixi (shared store) Report parser: fastp JSON Checkpoint: .oxo-flow/checkpoint.json
🤖 Unified PipelineGenAgent + Orchestrator Loop oxo-flow-ai
# Generate a validated pipeline with the Scientist Team profile $ oxo-flow template "paired tumor-normal WES with Mutect2 and Picard QC" \ --ai --ai-team-profile full -o somatic.oxoflow [1/4 Curator Brief] Extracted study design: paired somatic ([[pairs]]), GRCh38 reference, 4 rules, per-rule conda/docker pins. [2/4 Knowledge Base] Pinned verified tools from embedded KB: • fastp 1.3.0 (bioconda) • bwa-mem2 2.2.1 (biocontainers) • gatk4 4.6.2.0 (bioconda) [3/4 Orchestrator] Compiled candidate TOML → ran deterministic engine validator + DAG cycle/wildcard check → 0 errors [4/4 Senior Review] Verified resource ceilings & temporary=true intermediates. Saved session to .oxo-flow/ai_sessions/ # Auto-diagnose & self-heal runtime tool failures $ oxo-flow run somatic.oxoflow --ai-recover --ai-max-retries 3
🧠 5-Layer AI Architecture — Zero Overhead When Off Claude · OpenAI · DeepSeek · Ollama
# Enable across ALL CLI commands with 2 lines in .oxoflow: [ai] enabled = true team_profile = "full" # or default "compact"

Every surface — CLI template --ai, POST /api/ai/translate, and Web Copilot chat — runs the exact same PipelineGenAgent through the Orchestrator tool-calling and compiler-validation feedback loop.

Anti-Hallucination Guardrails Eval Benchmarked
Embedded Bioconda + bio.tools + Parabricks KB Compiler-in-the-loop validation MCP Server Bearer Auth SSRF Guard
🛡️ Runtime Disk-Pressure Guard & Protected Data Trees Runtime storage safety
# 1. Run with cryptographic output provenance & scratch redirection $ OXO_FLOW_TMPDIR=/scratch/tmp oxo-flow run wgs.oxoflow -j 16 --provenance ✓ Protected tree baseline verified: raw/ (48 files, 0 drift) ✓ Injected TMPDIR/TMP/TEMP = /scratch/tmp into all rule shells & hooks ⚠ Disk free dropped below min_free_disk (42G < 50G) → entering RECLAIM ✓ Cascaded tombstones for completed `temporary = true` rules (+94G freed) ✓ Resumed NORMAL scheduling · 12/12 rules succeeded # 2. Audit checkpoint sha256 output checksums at any time $ oxo-flow provenance verify .oxo-flow/checkpoint.json --json { "command": "provenance", "verify": { "matched": 36, "mismatched": 0, "missing": 0 } }
📉 Automatic 4-Stage Disk-Pressure State Machine No More Full-Scratch Crashes
1. NORMAL Free disk ≥ min_free_disk; full parallel dispatch.
2. RECLAIM Tombstones consumed temporary = true outputs.
3. HOLD Parks new dispatches while active rules finish.
4. ABORT Clean checkpoint save after grace rounds.
Regulated & Clinical Genomics Ready IQ / OQ / PQ
Protected data tree fingerprinter SHA-256 checkpoint provenance Shared OXO_FLOW_ENV_STORE + env gc
🌐 Three Deployment Modes + HPC Cluster Lifecycle Single Binary
# Mode 1: Personal workstation (SQLite, loopback, zero-config) $ oxo-flow serve --open # Mode 2: Multi-user Team server (RBAC + ORCID OAuth2 + AES-256-GCM keys) $ OXO_FLOW_PUBLIC_URL=https://flows.lab.org \ oxo-flow serve --mode team --host 0.0.0.0 -p 8080 # Mode 3: HPC Login-Node Submit Panel (auto-probes SLURM/PBS/SGE/LSF) $ oxo-flow serve --mode hpc # Direct CLI Cluster Submission & Live Queue Monitoring $ oxo-flow cluster submit wgs.oxoflow --backend slurm --queue gpu -o jobs/ $ oxo-flow cluster status --backend slurm
🖥️ 100+ OpenAPI 3.1 Endpoints Across 9 Domains React 19 + Axum + React Flow

Includes interactive DAG editor (/editor), live SSE Run Monitor (/runs/:id), AI Copilot Chat (/chat), Cluster Dashboard (/clusters), Audit Log (/audit), Embedded Swagger Docs (/docs), and oxo+https:// Share URLs.

CLI ↔ Web Run Staging Parity GET /api/openapi.json
include_files staging config_overrides (--arg) /protected-drift preview /provenance-verify audit
100%
Safe Rust across workspace crates (#![forbid(unsafe_code)])全工作区安全 Rust (#![forbid(unsafe_code)])
8backends
conda, mamba, pixi, docker, singularity, venv, system, modulesconda / mamba / pixi / docker / singularity / venv / system / modules
35 cmds
Complete CLI lifecycle + strict stdout/stderr separation覆盖完整生命周期的 36 个 CLI 子命令
100+APIs
OpenAPI 3.1 REST endpoints across 9 modular domains9 大领域模块、100+ OpenAPI 3.1 接口
16+Hub
CI-tested Gallery pipelines + community nf-core/Snakemake ports16 个内置 CI 验证流程 + 社区移植中心
3,400+
Automated unit, integration, CSP, schema-drift & AI eval tests3,400+ 单元、集成、安全与 AI 评测测试
01 · CORE ARCHITECTURAL INNOVATIONS

Engineered for deterministic genomics at scale. 专为大规模确定性生物信息学计算打造的六大创新。

Traditional workflow managers bolt environments, UIs, and ad-hoc scripts onto slow interpreters. oxo-flow unifies compilation, graph scheduling, AI synthesis, and storage safety in a single memory-safe binary. 传统工作流引擎往往在解释型语言之上拼凑环境管理、Web 界面与脚本。oxo-flow 在单一内存安全二进制中融合了静态图编译、资源调度、AI 验证闭环与存储安全守护。

🔀 Petgraph DAG Engine · Wildcards · [[pairs]] · [[transform]]

Target-Aware DAG Compilation & High-Order Bio-Operators 目标感知的 DAG 编译器与高阶生信算子

Compiles .oxoflow TOML into a verified Directed Acyclic Graph with cycle detection, regex-constrained wildcards ({sample}, {chr}), conditional branches (when), unified scatter-gather ([[transform]]), and first-class tumor–normal somatic pairing ([[pairs]]) plus multi-cohort sample groups. Use -t to compute and run the minimal subgraph required for any target deliverable. 将 .oxoflow 编译为经过环检测的有向无环图,原生支持正则约束通配符({sample}、{chr})、条件规则(when)、统一 Scatter-Gather 算子([[transform]])、肿瘤-对照配对([[pairs]])及多队列分组聚合。支持 -t 最小子图按需执行与 nf-metro 地铁图导出。

# Export Graphviz DOT or Seqera nf-metro transit map $ oxo-flow graph pipeline.oxoflow -f metro -o pipeline.mmd $ oxo-flow run pipeline.oxoflow -t vcf/cohort.filtered.vcf.gz -j 16
🤖 Unified AI Companion · Scientist Team

Compiler-in-the-Loop AI & Scientist Team 编译器闭环的 AI 伴侣与科学家团队

A single [ai] enabled = true line activates AI across template, validate, lint, dry-run, debug, and run --ai-recover. The shared PipelineGenAgent + Orchestrator validates every generated workflow against the real engine and an embedded Bioconda / bio.tools / Parabricks knowledge base. Opt into --ai-team-profile full for Curator brief + independent Reviewer verification. 只需一行 [ai] enabled = true 即可在全部 CLI 与 Web 端启用 AI。共享的 PipelineGenAgent 与 Orchestrator 将每次生成与真实引擎验证器及内置生信知识库(Bioconda / bio.tools / Parabricks)闭环校验;--ai-team-profile full 更可启动 Curator + Reviewer 多智能体审查。

$ oxo-flow template "scRNA-seq CellRanger + Seurat" \ --ai --ai-team-profile full
🛡️ Storage Safety · Provenance

Disk-Pressure Guard, Scratch Isolation & Protected Trees 磁盘压力自动守护、全局临时目录与受保护数据树

Bioinformatics runs often crash at 95% completion when BAM/FASTQ intermediates exhaust scratch storage. oxo-flow’s [engine] section enforces a live Normal → Reclaim → Hold → Abort ladder that cascades tombstones up temporary = true rules automatically, injects global tmp_dir (TMPDIR/TMP/TEMP), guards raw data via protected_data_trees, and records cryptographic sha256 provenance in checkpoint.json. 生信流程常因中间 BAM/FASTQ 撑爆磁盘而在终点前崩溃。[engine] 原生提供 Normal → Reclaim → Hold → Abort 四级磁盘压力状态机,自动级联清理 temporary = true 规则产物,注入全局 tmp_dir,通过 protected_data_trees 防止原始数据被篡改,并记录 sha256 断点溯源。

📦 8 Environment Backends · OXO_FLOW_ENV_STORE

Per-Rule Isolation + Content-Addressed Shared Env Store 单规则级 8 后端环境隔离与内容寻址共享存储

Mix and match 8 backends within the same workflow. Set OXO_FLOW_ENV_STORE to deduplicate pixi and venv environments across all projects by content hash (with symlink materialization and checkpoint-aware oxo-flow env gc), or export the entire pipeline into a multi-stage Docker or rootless Singularity image. 支持在同一流程中按规则混用 8 种环境后端。设置 OXO_FLOW_ENV_STORE 即可按内容哈希跨项目去重共享 pixi 与 venv 环境(支持 oxo-flow env gc 断点感知垃圾回收),并内置国内镜像加速与 Docker/Singularity 一键导出。

conda mamba pixi docker singularity venv system HPC modules
📊 Execution-Truth Reporting · Domain QC

Multi-Format Reports & Native Bioinformatics QC Parsers 基于执行真相的多格式报告与原生生信 QC 解析

Generates self-contained HTML, JSON, Markdown, and PDF reports directly from verified checkpoint data — plus native output parsers for fastp, samtools flagstat, STAR, featureCounts, bcftools, and kraken2, and one-flag export to R-ready TSV tables (--r-data). 直接从断点真相数据生成 HTML、JSON、Markdown 与 PDF 报告,内置 fastp、flagstat、STAR、featureCounts、bcftools、kraken2 原生指标解析器,并支持 --r-data 导出 R 语言友好的 TSV 矩阵。

🧩 VS Code IDE · React 19 Web Studio · Native Desktop

First-Party VS Code Extension, Web SPA & Native Desktop App 官方 VS Code IDE 插件、React 19 Web 工作台与原生桌面应用

Author in VS Code / Cursor / Windsurf with traitome.oxo-flow (schema-driven autocomplete, hover docs from oxoflow-v1.schema.json, live CLI lint/validate diagnostics, canonical formatting, one-click run/dry-run/graph). Or launch the visual React Flow DAG studio via oxo-flow serve or the native Wry/Tao desktop app on macOS and Linux. 使用官方 traitome.oxo-flow 插件在 VS Code / Cursor 中获得基于 JSON Schema 的智能补全、悬停文档、实时诊断与一键运行;或通过 oxo-flow serve 及 macOS / Linux 原生桌面安装包启动可视化 React Flow 工作流工作台。

01B · ONE ENGINE, TWO AUDIENCES

Built for people and agents — same engine, same guarantees. 为人类与 Agent 而生——同一引擎,同一套保证。

An agent calls the exact commands a person runs and gets the same evidence back. There is no shadow API and no AI-only execution path: every generation, repair, and run passes the engine's own validate → dry-run → run → provenance gates. Agent 调用与人类完全相同的命令,拿回同样可核对的结果。没有影子 API,也没有只给 AI 的执行路径:每次生成、修复与运行都必须通过引擎自身的 validate → dry-run → run → provenance 门禁。

🧑‍🔬 For people

Readable gates, resumable runs, review-grade evidence 可读的门禁、可恢复的运行、可用于评审的证据

  • One binary, four first-party surfaces: CLI, Web Studio, VS Code extension, native desktop.单一二进制,四种官方界面:CLI、Web 工作台、VS Code 插件、原生桌面端。
  • Static gates before compute, with stable E/W codes and remediation suggestions.计算前先过静态门禁,诊断带稳定 E/W 编码与修复建议。
  • Checkpoint resume, SHA-256 provenance receipts, and HTML/JSON/Markdown/PDF reports from execution truth.断点续跑、SHA-256 溯源凭据,以及基于真实执行数据的 HTML/JSON/Markdown/PDF 报告。
  • Disk-pressure guard, scratch isolation, 8 isolated environment backends.磁盘压力守护、临时目录隔离、8 种环境后端隔离。
🤖 For agents

A machine contract the engine itself enforces 由引擎自身强制的机器契约

  • oxo-flow agent-docs scaffolds AGENTS.md, CLAUDE.md, and the full SKILL.md manual — embedded in the binary, offline-capable.oxo-flow agent-docs 生成 AGENTS.md、CLAUDE.md 与完整 SKILL.md 手册——内嵌于二进制,离线可用。
  • Every gate has --json with stable codes and exit codes; errors use {code, message, detail, suggestion}.每个门禁都支持 --json(稳定编码与退出码);错误统一为 {code, message, detail, suggestion}。
  • AI output is compiler-validated: generated TOML must pass the engine's schema and static gates.AI 产出需经编译器校验:生成的 TOML 必须通过引擎 schema 与静态门禁。
  • Data-safety contract in the skill, protected_data_trees drift detection, and a hard OXO_FLOW_AI_PROVIDER=disabled pin.技能内置数据安全契约、protected_data_trees 漂移检测,以及不可被持久配置覆盖的 OXO_FLOW_AI_PROVIDER=disabled 硬开关。
02 · HEAD-TO-HEAD ARCHITECTURAL COMPARISON

Why teams migrate to oxo-flow. 为什么生信团队正在转向 oxo-flow?

A side-by-side look at how a compiled Rust-native architecture compares against Python- and JVM-based pipeline engines. 全面对比 Rust 原生编译架构与传统基于 Python 或 JVM 的工作流引擎在性能、安全、环境与智能化上的代际差异。

Dimension ⚡ oxo-flow (Rust 2024) Snakemake Nextflow
Core Runtime & Memory Safety Compiled native binary, #![forbid(unsafe_code)] across workspace crates Python interpreter runtime Groovy on JVM runtime
Startup & DAG Resolution Instant millisecond startup; petgraph verified on 1,000-rule stress tests Python import + solver latency JVM warm-up & daemon overhead
Workflow Definition Format Declarative, schema-validated TOML (.oxoflow) + modular [[include]] Snakefile (Python-mixed DSL) Nextflow Groovy DSL2
Per-Rule Environment Backends 8 backends: conda, mamba, pixi, docker, singularity, venv, system, HPC modules + shared content store (OXO_FLOW_ENV_STORE) conda, mamba, singularity, docker, envmodules conda, mamba, docker, singularity, spack, modules
Built-in AI Companion Native oxo-flow-ai: NL generation, compiler-in-the-loop validation, Scientist Team review, auto error recovery (--ai-recover) Not built-in External / Seqera AI
Storage & Data Protection Runtime Disk-Pressure Guard (Normal→Reclaim→Hold→Abort), protected_data_trees, global tmp_dir Manual temp() markers only Workdir cleanup on completion
Provenance & Regulated Audit Built-in --provenance SHA-256 checkpoint verification + IQ/OQ/PQ validation protocol Archive / report flags Trace / Tower lineage
Web Studio, Desktop & REST API Embedded in binary (oxo-flow serve): React 19 SPA, 100+ OpenAPI 3.1 endpoints, Personal / Team / HPC modes + native Desktop app External panoptes / community UIs Commercial Seqera Platform (Tower)
Reporting & QC Parsing Built-in HTML/JSON/MD/PDF reports + native fastp, STAR, flagstat, featureCounts, bcftools, kraken2 parsers + --r-data TSV Requires external MultiQC / Quarto Execution timeline HTML
First-Party IDE Extension traitome.oxo-flow (VS Code Marketplace & Open VSX) synced to JSON Schema on every release Syntax highlighting plugins Nextflow VS Code extension
03 · COMPLETE LIFECYCLE TOOLING — 36 CLI SUBCOMMANDS & 9 WEB DOMAINS

Every command & endpoint, documented and linked. 36 个 CLI 子命令与 9 大 Web API 领域,一键直达权威文档。

Filter all 36 subcommands of the oxo-flow binary below — click any card to open its full command reference with flags, JSON schemas, and examples. 按分类筛选全部 36 个 oxo-flow CLI 子命令 —— 点击任意卡片即可跳转至对应的详细命令参考手册。

runexec
Execute DAG workflows in parallel with resource scheduling, disk guard & --provenance.
resumeexec
Resume interrupted workflows from checkpoint.json skipping completed rules.
dry-runexec
Preview execution plan (alias: plan), resource estimates, and AI safety audit.
testexec
Execute workflow test suites and fixture assertions deterministically.
batchexec
Run parameterized batch executions across cohorts or sample sheets.
initdev
Scaffold a new .oxoflow project directory with standard structure.
agent-docsdev
Scaffold AGENTS.md, CLAUDE.md, and the full SKILL.md operating manual for AI agents.
templatedev
Generate pipelines from built-in templates or natural language (--ai / --from-url).
importdev
Transpile Nextflow DSL2 and Snakemake pipelines into native typed .oxoflow workflows.
validatedev
Validate TOML syntax, DAG acyclicity, wildcards, and semantic AI checks.
lintdev
Static best-practice analyzer with reproducibility checks; --ai adds semantic linting.
formatdev
Canonical formatter for .oxoflow files (powers VS Code Format Document).
debugdev
Inspect expanded shell commands, wildcards, and AI failure diagnosis.
graphinspect
Export DAG as Graphviz DOT, tree, Mermaid, or Seqera nf-metro transit map.
reportinspect
Generate HTML, JSON, Markdown, or PDF reports + domain QC & --r-data TSV.
provenanceinspect
Export and cryptographically verify sha256 output checksums in checkpoint.json.
verify-datainspect
Verify a raw data tree against a vendor md5sum sidecar before runs.
statusinspect
Inspect per-rule checkpoint completion state and output freshness.
waitinspect
Block until a run in a workdir finishes; exit 0 ok, 1 failed, 3 timeout, 4 no run.
infoinspect
Summarize workflow metadata, rule counts, environments, and resource totals.
configinspect
Inspect resolved configuration, --profile supplements, and --arg overrides.
diffinspect
Compare two .oxoflow workflows structurally across rules, envs, and params.
schemainspect
Emit the canonical oxoflow-v1 JSON Schema for IDE & CI validation.
envenv
Create, list, inspect, and garbage-collect (env gc) isolated rule environments.
runsenv
Reclaim stale per-run workspace trees (runs gc) behind checkpoint-ref and age guards.
exportenv
Package workflows into multi-stage Dockerfiles or rootless Singularity recipes.
cleanenv
Safely remove generated outputs, logs, or temporary intermediates.
touchenv
Update output timestamps and synchronize checkpoint state without re-running.
servedeploy
Start the embedded Axum REST API & React 19 Web Studio (personal / team / hpc).
clusterdeploy
Submit, monitor, and cancel jobs on SLURM, PBS, SGE, and LSF schedulers.
publishdeploy
Package and publish reproducible workflow bundles with checksums.
pulldeploy
Download and verify remote workflow bundles or shared oxo+https:// pipelines.
knowledgeai
Query embedded Bioconda tools, nf-core modules, skills, docs & pipeline graph.
aiai
Configure AI providers, inspect active models, and test endpoint connectivity.
completionssystem
Generate shell completions for bash, zsh, fish, PowerShell, and elvish.
licensesystem
Inspect crate licensing and verify commercial/academic web license status.
🌐 9 Web API Domains · 100+ Endpoints · GET /api/openapi.json

AI-Native Axum REST API & Real-Time SSE Event Stream AI 原生 Axum REST API 与实时 SSE 事件流(9 大领域模块)

domains/workflow · domains/execution · domains/dag · domains/ai · domains/auth · domains/collaboration · domains/observability · domains/chat · domains/clusters

05 · COMPLETE NAVIGATION & IMPORTANT LINKS ATLAS

Every key resource, explained and indexed. 核心文档、工具注册表与合规协议全景导航中心。

Start with your role below, or browse the complete 6-domain resource directory covering tutorials, operational guides, engine specifications, AI/Web APIs, package registries, and regulated compliance protocols. 可按您的角色快速选择起点,或在下方的六大资源矩阵中查阅全部教程、操作指南、架构规范、AI/Web 接口、软件包镜像与合规验证文档。

FOR NEW USERS & RESEARCHERS

First 15 Minutes with oxo-flow新手 15 分钟极速入门

Install the single binary, scaffold a project, run your first DAG, and generate an HTML report.安装单一二进制文件,初始化项目,运行第一个并行 DAG 并生成 HTML 报告。

FOR BIOINFORMATICIANS & PIPELINE AUTHORS

Authoring Production Pipelines编写生产级生物信息学管线

Master wildcards, tumor–normal [[pairs]], cohort groups, custom Python/R scripts, and VS Code IDE tooling.掌握通配符、肿瘤-对照配对、队列分组、内嵌 Python/R 脚本与 VS Code 插件。

FOR HPC, CLOUD & CORE FACILITY ADMINS

Cluster Scheduling, Security & ComplianceHPC 集群调度、多用户部署与合规审计

Deploy Team/HPC web modes, configure SLURM/PBS/SGE/LSF, enable shared env stores, and run IQ/OQ/PQ validation.部署 Team/HPC Web 服务,配置 SLURM/PBS/SGE/LSF 集群、共享环境存储与 IQ/OQ/PQ 验证。

06 · ACADEMIC CITATIONS, LICENSING & CONTRIBUTORS

Cite oxo-flow in your research. 在您的学术研究中引用 oxo-flow。

If oxo-flow or its companion ecosystem contributes to your research publication, please cite our papers below. 如果 oxo-flow 对您的科研工作有所帮助,请引用以下论文(支持一键复制 BibTeX)。

PRIMARY CITATION · BIORXIV (2026)

oxo-flow: compiled, memory-safe bioinformatics workflow orchestration

Shixiang Wang · bioRxiv, 2026
COMPANION ECOSYSTEM · BIOINFORMATICS ADVANCES (2026)

BLIT: an R package for seamless integration of command-line bioinformatics tool universe

Jia Ding, Yun Peng, Ruochen Wei, Boquan Wang, Jian-Guo Zhou, Shixiang Wang · Bioinformatics Advances, Vol. 6, Issue 1, 2026, vbag088
⚖️ Open-Core & Academic-Free Licensing ⚖️ 开放核心与学术免费授权模式

Apache-2.0 Core, CLI, AI & Desktop · Academic-Free Web Apache-2.0 核心/CLI/AI/桌面端 · 学术免费 Web

oxo-flow-core, oxo-flow-ai, oxo-flow-cli, and oxo-flow-desktop are 100% free and open-source under the Apache License 2.0. oxo-flow-web is dual-licensed: free for academic, non-profit, and personal research under LICENSE-ACADEMIC, with LICENSE-COMMERCIAL for commercial deployments. oxo-flow-core、oxo-flow-ai、oxo-flow-cli 与 oxo-flow-desktop 遵循 Apache 2.0 许可证,完全开源免费。oxo-flow-web 采用双重授权:学术机构、非营利组织与个人研究在 LICENSE-ACADEMIC 下免费使用,企业商业部署采用 LICENSE-COMMERCIAL。

🤝 Contributors & Governance 🤝 贡献者与工程治理

Built by Traitome & the Open-Source Community 由 Traitome 团队与开源社区共同构建

Every pull request is gated by make ci (rustfmt, clippy -D warnings, 3,400+ tests, cargo-audit, cargo-deny, gitleaks, schema drift, and AI eval lint). 每个 Pull Request 均由 make ci 严格门禁把关(包括 rustfmt、clippy -D warnings、3,400+ 自动化测试、cargo-audit、cargo-deny、gitleaks、模式漂移与 AI 评测门禁)。