Busabase
Busabase 是一个开源(MIT)数据库与 AI 智能体工作空间——将数据、文档、技能和应用集成于一体。您可以本地运行、在网络内自托管或使用 Busabase Cloud,从而开启新时代。
AI 翻译
尚未收集仓库信息。
尚未检测到更新。GitHub 发布版本后会显示在这里。
制作者的介绍仅有上面的一句话。 更多详情请阅读仓库 README。 仓库 README ↗
正在加载完整 README。
查看完整 README ↗A browser-based Hi-C contact map viewer and genome assembly curation tool. It reads native .pretext files, renders them with WebGL2, and lets you curate an assembly into chromosomes, with nothing to install.
Try it now · Field guide to genome curation · Report an issue
Open the app and click Load Example to explore a real koala assembly, or Start Tutorial for a guided walkthrough.
Contact map with contig sidebar, misassembly badges, annotation tracks, and analysis overlays.
OpenPretext is a browser-based alternative to PretextView, the Wellcome Sanger Institute desktop tool that genome assembly teams (Darwin Tree of Life, Vertebrate Genomes Project, Earth BioGenome Project) use to inspect Hi-C contact maps and arrange contigs into chromosomes. OpenPretext runs anywhere a browser does, needs no install, and adds scripting, automated curation, 3D genomics analysis, and an optional machine-learning enhancement server.
New to genome curation? Start with the field guide. It teaches you to read a contact map and recognize the common misassemblies before you touch the tool.
Prerequisites: Node.js 22 or later.
git clone https://github.com/shandley/openpretext.git
cd openpretext
npm install
npm run dev
Open http://localhost:3000, then:
.pretext file from your computer.OpenPretext needs a local HTTP server; opening the files over file:// will not work (CORS and ES module loading). Use npm run dev while developing, or npm run build then npm run preview for a production build.
WebGL2 contact map with tile-based level-of-detail, an LRU cache, and background decompression. Six color maps with keyboard cycling, adjustable gamma, a contig grid and edge labels, a click-to-navigate minimap, scaffold color bands, waypoint markers, annotation track overlays (line, heatmap, marker), and a comparison overlay of original versus curated boundaries. An observed/expected (O/E) view toggle divides out the distance-decay so long-range structure stands out (overview only; detail tiles stay raw at high zoom). A before/after view places the map as loaded beside the current curated arrangement.
Cut, join, invert, and move contigs with full undo and redo. Drag-and-drop reordering, a click / shift-range / ctrl-toggle selection system, contig exclusion (hide from export without deleting), scaffold painting for chromosome assignment, waypoint markers, batch operations (select by name or size, batch cut / join / invert, sort by length), and meta tags to classify contigs as haplotig, contaminant, unlocalised, or sex chromosome. Tags and exclusions follow a contig when it is cut; a join of an excluded contig with an included one is refused, because either result would change what is exported.
A common recipe: run Auto Cut first to break misassemblies, then Auto Sort to group and orient the fragments. Every operation is individually undoable.
Vision-based analysis through the Anthropic Messages API. It captures the current map, builds assembly context from contig ordering, metrics, and scaffold assignments, and returns runnable DSL command suggestions with one-click Run buttons. Ships with eight prompt strategies (general, inversion detection, scaffold assignment, fragmented assemblies, micro-chromosomes, analysis-guided, haplotig detection, telomere-aware), a custom-strategy editor, JSON export and import, and one-click browsing of community strategies from the panel. Your API key is stored locally and sent only to Anthropic.
A suite of Hi-C metrics, all computed in a background Web Worker and exportable as BedGraph or TSV:
以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗