AI agent-assisted bioinformatics analysis platform
Ask in plain language — the agent runs the analysis, checks it and returns figures, statistics and a report.
Built for how a PI works
Deadlines are close; the figures are still in a student’s notebook.
With EuchroGeneEvery figure is registered with methods and statistics; the Writing agent drafts from them.
Lab data is spread over laptops and old drives.
With EuchroGeneEach study gets dated folders — design, data, analysis, figures, manuscript.
A student graduates; nobody can rerun the analysis.
With EuchroGeneScript, versions and parameters are saved beside each figure.
Too many papers, too little time.
With EuchroGeneThe lab library answers with page citations and quizzes you on a paper.
Records scatter; corrections vanish.
With EuchroGeneLabNote entries are signed, hash-locked and retracted only with a reason.
Progress updates arrive late, without numbers.
With EuchroGeneMembers report to the PI from what was actually recorded.
Say what you need. The agent does it — and shows its work.
It works in your lab’s Jupyter and registers every result with what you need to trust and rerun it.
-
Request
Ask in plain words.
-
Plan
Data, method, check and figure — shown first.
-
Run
Code runs in your Jupyter.
-
Check & fix
Failed check → fix → rerun.
-
Result & record
Figure with statistics, Methods and script.
Pipeline Dev. loop
Run, check, fix — stops on a repeated error or the attempt limit.
Every figure traceable
Statistics, Methods and a copy of the exact script (SHA-256).
Planned first
Analysis plan, n and power fixed at design; drift is flagged.
Bioinformatics built in
RNA-seq, variants, BLAST, alignments, CRISPR, primers, cloning.
Your compute
Lab Jupyter or your own GPU server; switches AI at a usage limit.
Into papers and slides
Results flow to the manuscript, tables, Word and PowerPoint.
You stay in control: plans first, deletions confirmed, numbers checked, every figure traceable.
Analyse, plan, write, present, read, record
Four agents per project — Design, Analysis, Writing, Papers — plus a LabNote. You decide every open question.
Analyses the agent runs, checks and explains — and anyone can rerun
Ask for a step; the agent writes the code in your Jupyter, runs it, checks and fixes it, and explains the result.
- NewPipeline Dev. loop: run, check, fix, rerun
- Figures registered with Methods, statistics and their script
- RNA-seq, variants, amplicons, metagenomes, assemblies; BLAST, CRISPR, primers, cloning, JBrowse2
A study design your whole team can follow
The Design agent plans from your data, references and notes — with controls and n sized by power.
- Flowchart and timeline, versioned
- Every step becomes an assigned task
- Warnings for missing controls, n or analysis plan
Drafts built on your results, not invented ones
Drafts from registered figures and verified references only, as tracked changes you accept.
- References checked by DOI / PMID; unverifiable ones left as [REF]
- Journal checks, review simulation, reviewer responses
- Grants, progress reports and reports to the PI
Slides and Word documents in the format you need
Make PPT / Make Word builds slides or a document from your figures.
- Purpose, depth, audience, theme — or your own template
- Report, summary, proposal, protocol, review — with PDF
- Conference posters as one editable slide
A lab library that reads with you
A shared library searched by meaning; answers cite pages.
- Study a paper at three depths, then a quiz
- Study log, bookshelves, alerts
- One-click abstract translation
A lab notebook that holds up
Agents and members record results, decisions, methods and deviations.
- Sign and witness with a password; SHA-256 locked
- Corrections stay visible as retractions
- Templates, weekly and lab summaries
The whole lab in one place — even on your phone
Files, tasks, calendars and messages next to the research.
- DataShare with previews and versions
- Projects, calendars, messages
- Mobile app with push notifications
Every project at a glance
Lab overview
Stage, next step, deadline and unsigned results of every project.
Deadlines
D-n on each card; due within 14 days moves to the top.
Weekly Lab summary
Key points, deviations and overdue steps from the LabNote.
Lab roles
PI, researcher and staff; personal notes stay their owner’s.
Funded grants
One project per Specific Aim; progress reports from their records.
Nothing signed for you
Agents never sign LabNote entries or mark papers published.
Claude or Codex — and it keeps working
Connect your own subscription, or the PI adds a lab key.
NewSwitch at a limit
At a usage limit, the other AI answers and hands back after the reset.
NewPoint at what you mean
Tick a figure, section, reference or file — the agent works on exactly that.
Checked, not just generated
A reviewer pass checks numbers and claims against tool output.
Your language
Ask in any language; manuscripts are written in American English.
Pick the model for each engine; every chat can switch between Claude and Codex.
Sized for real research
Your data stays your lab’s
Own server and accounts per lab — nothing shared.
Per-lab isolation
Dedicated server and database, role-based access.
Secure sign-in
Two-factor login, rate limits, idle sign-out, audit log.
Privacy & backups
Export and deletion requests (GDPR / PIPA), nightly backups.
What the AI sees
Only what a question needs, sent to the provider you connected.
Three steps to start
Tell us about your lab
Field, team size, what you want to see — we demo it.
We set up your server
Accounts and roles; setup and updates included.
Your team signs in
Browser, JupyterLab or phone — then ask the Design agent.
One predictable yearly price
The dedicated server, the platform software, updates, maintenance and email support in one annual rate — no per-hour metering and no surprise fees. AI usage runs on your own Claude / ChatGPT subscription or a lab API key, billed by that provider.
- 32 vCPU · 256 GB RAM · 4.3 TB, Ubuntu 22.04 LTS
- Research projects, Papers, LabNote and the AI agent
- BLAST, sequence design tools, JBrowse2 and JupyterLab
- DataShare, Project, calendar, messages and the mobile app
- Managed setup and updates
Frequently asked questions
Can I trust an analysis the agent ran?
Every figure the agent registers carries its statistics, Methods, an interpretation and a copy of the exact script and environment file (with its SHA-256), so anyone in the lab can see how it was made and rerun it. The agent shows its plan first, checks each run against acceptance criteria, asks before steps that could delete your data, and a reviewer pass checks the numbers in its answers against the tools’ output. You decide what goes into the paper.
Who owns our data, and where is it stored?
Your lab does. Each lab runs on its own dedicated server with its own database and accounts, and nothing is shared with other labs. Members can export their data, and account deletion requests (GDPR / PIPA) are handled on the platform.
Is our unpublished data sent to an AI company?
Only when someone asks the agent, and only the parts that question needs — sent to the provider you connected (Anthropic for Claude, OpenAI for Codex) under your own account or the lab key. Files, notebooks and records otherwise stay on your lab’s server.
Which AI does it use, and do we need our own subscription?
Claude (by Anthropic) or Codex (by OpenAI). Each member can connect a personal Claude subscription or ChatGPT account, or a PI can add a lab shared key. With both connected, each chat picks its engine and the other takes over when one reaches its usage limit.
Can the AI make up numbers or references?
The Writing agent takes numbers only from registered figures, their statistics and the LabNote; a missing analysis becomes a request to the Analysis agent instead of an invented value. References are verified by DOI / PMID, and anything it cannot verify is left as [REF] for you. A second reviewer pass checks each answer against the tools’ actual output.
What does it cost?
One yearly flat rate per lab covering the dedicated server, the software, updates, maintenance and email support. AI usage is billed separately by the provider you connect. Request a quote and we will send pricing for your team.
Can my students and postdocs use it, and what can they change?
Yes — every member gets an account with a PI, researcher or staff role. Everyone can read the lab’s research projects; owners, project members and PIs change them. Personal LabNote entries can be changed only by their owner.
Does it work in Korean?
The interface is in English. You can ask the agents in Korean or any language and they answer in yours; a translation view shows a manuscript in your language. Manuscripts are written in American English, and a Korean NRF proposal or a report can be written in Korean.
We already have servers or a campus cluster. Can we use them?
Yes. Your lab’s own servers, such as GPU machines, can be connected to Jupyter and switched to with one click; the agent can run there too. File Transfer copies data between your Jupyter and another server over SSH, with a key or just the server password.
Do we need IT staff or any installation?
No. Everything runs in the browser, in JupyterLab or in the phone app, and we set up, update and maintain the server.
See it with your own lab’s questions
Tell us what your lab works on — we will show you the platform on a demo and send specifications and pricing for your team.