Pan.bio
Bioinformatics workflow automation platform based on nf-core pipelines.
- bioWorkflows: An automated workflow engine that runs validated nf-core pipelines (Sarek variant calling, RNA-seq, scRNA-seq, Amplicon-seq, Metagenomics, etc.) without coding. Ensures version control and reproducibility from FASTQ upload to VCF, BAM, and clinical reports. Supports direct import of user's Nextflow workflows from GitHub. Adaptive compute (automatically allocates high CPU for alignment and variant calling, and minimal resources for lightweight tasks). Processes WGS in 8-20 hours and RNA-seq in ~2 hours. Clinical workflow with 2-caller consensus achieves SNV precision ≥94%, SNV recall ~99%, and INDEL recall 100%. - bioNotebooks: Jupyter-style interactive analysis environment. Pre-installed with Python 3.11 (pandas, NumPy, Plotly, matplotlib) + R (ggplot2) + CLI tools (samtools, bedtools). Based on the bionl kernel. Allows direct access to GEO/SRA/IPG public datasets using accession numbers. - bioVAIC: AI-based variant interpretation and classification. Classification based on ACMG/AMP, CanVIG-UK, ACGS, ClinGen, and gene-specific guidelines (BRCA1, BRCA2, TP53, ATM, APC, MMR). Automatically collects evidence from ClinVar, gnomAD, OMIM, PubMed, BayesDel, REVEL, and SpliceAI. Supports VCF, HGVS, and single variant input. Dual AI support: VAIC Agent (classification guide) + Research Agent (literature and database search). Automatically generates clinical reports, while the final classification remains at the discretion of the clinician. - bioCohorts: Trusted Research Environment (TRE) marketplace specializing in underrepresented populations (MENA, East Asia). Contains 12,400 genomic datasets, 8,200 clinical records, 3,100 images, and 1,800 proteomics datasets. Supports WGS/WES, RNA-seq, clinical EHR, images (CT, MRI, X-ray), and patient metadata (demographics, longitudinal tracking). Analysis is performed within an isolated, secure workspace without transferring raw data. License agreement model based on patient, year, and data type. - BioMind AI Copilot: An AI layer integrated into all products. The BioAgents layer (domain-specific agents + shared skill and tool library) automatically generates and executes Python/R code from natural language. Connects to external tools via an MCP (Model Context Protocol) server. Cohort Discovery Agent — converts natural language queries into structured filters and transparently discloses gaps. - Bion Multi-Agent System: The Orchestrator Agent coordinates three specialized agents: Code Generator, Filesystem, and Backend. Processes the entire workflow from Import → Clean → Analyze → Visualize → Iterate within the notebook. - Data Sovereignty Architecture: Deployed in GCC region (Saudi Dammam GCP), US-East/US-West GCP. Prevents cross-border data transfer at the architectural level (not just through policy). AI processing is performed within the region (all BioMind LLM calls are routed to region-specific infrastructure). - Security and Compliance: HIPAA, GDPR, SOC 2 Type II, ISO 27001 — enforced at the infrastructure level. Row-level tenant isolation (not possible due to cross-tenant query architecture). Immutable audit logs (records all AI tool calls, LLM calls, and data access, compliant with ISO 15189, CAP, and HIPAA). RBAC + human approval required for clinical write operations. Patient identifiers are server-side only (not exposed in the browser), and only de-identified metadata is used for cohort operations.
💻 System Requirements
Not applicable (server-side computation, GPU infrastructure provided by Pan.bio)
Not applicable (cloud storage, allocated based on plan)
⚡ Installation
### 4-1. Quick Start
```
# Web SaaS — No installation required
# 1. Create an account at https://lab.pan.bio/register (free trial, no credit card required)
# 2. After logging in, select bioWorkflows / bioNotebooks / bioVAIC / bioCohorts
# 3. Upload FASTQ files or access public data using GEO/SRA accession numbers
```
### 4-2. Detailed Installation
```
# Pan.bio is a fully managed cloud SaaS platform
# Local installation, Docker, pip install, etc., are not required
# Access paths:
# - Platform: https://lab.pan.bio/
# - Sign-up: https://lab.pan.bio/register
# - Login: https://lab.pan.bio/login
# - Demo reservation: https://calendar.app.google/cGEaRG5uJ21zUWUT7
# Enterprise / In-Country Deployment:
# - Separate inquiry required (deployment of dedicated infrastructure within the country)
# - Select GCC (Saudi Dammam), US-East/West region
# - Compliant with HIPAA, GDPR, SOC 2 Type 2, ISO 27001
```🧬 Bio Use Cases
WGS/WES Variant Analysis Pipeline
Upload FASTQ raw data to the Sarek pipeline in bioWorkflows to automatically perform alignment, variant calling, and annotation (WGS 8-20 hours, RNA-seq ~2 hours). 2-caller consensus-based SNV precision ≥94%, recall ~99%. Transfer the results to bioVAIC for variant classification based on ACMG/AMP guidelines — automatically collect evidence from ClinVar, gnomAD, OMIM, PubMed, BayesDel, REVEL, and SpliceAI, and then generate a clinical report. Complete tasks that previously took bioinformaticians several days in a CLI environment without coding.
Precision Medicine Research for MENA and East Asian Populations
Acquire reference data for Arab and East Asian populations, which are missing from global public databases (gnomAD, ClinVar), from the bioCohorts TRE Marketplace, including genomic data from 12,400+ individuals and clinical records from 8,200+ individuals. Analyze the data using BioMind AI in bioNotebooks with natural language queries such as "Compare the frequency of BRCA1 variants in this cohort with gnomAD." Ensure data sovereignty by analyzing data within an isolated workspace without transferring raw data. Utilize this data for precision medicine research to reduce the risk of false positives/negatives caused by European-biased reference data.
Biomarker Discovery for Pharmaceutical and Biotechnology Companies
A pharmaceutical company discovers a cohort of patients with a rare disease in the bioCohorts Marketplace using natural language search with Discovery Agent, reviews the aggregated metadata, and gains access to the analysis workspace through a licensing agreement (completed within weeks compared to several months for existing data access agreements). When performing RNA-seq differential expression analysis in bioNotebooks, BioMind AI automatically generates statistical code with natural language queries, enabling non-experts to perform exploratory analysis. Data ownership remains with the original provider, enabling compliant collaboration.
📝 Update Notes
No update notes yet.
🧪 Related Code of Life
No related Code of Life posts yet.