Loading
NIOME is a decentralized AI subnet that enables privacy-safe genomic intelligence by replacing real human genomes with high-fidelity synthetic genomic profiles
Neuron registration trends
Genome editing research faces a critical bottleneck: generating large-scale, diverse CRISPR datasets is expensive, time-consuming, and limited to a small number of gene targets. Existing datasets are biased toward common loci and lack the mutation diversity needed for robust model training, particularly for disease variants like those in the HBB gene relevant to sickle cell disease and beta-thalassemia.
Decentralized networks can address this by coordinating many independent contributors to generate synthetic experimental data at scale, each operating under the same biological validation constraints. This distributes the computational load while ensuring all outputs meet consistent quality standards through multi-stage evaluation.
NIOME implements this on Bittensor subnet 55 by having miners generate lists of CRISPR experimental designs (guide sequences, target positions, mutations, Cas systems) conforming to task specifications, while validators evaluate each submission through five progressive validation stages—from structural correctness to biophysical simulation to distributional fidelity against real CRISPR data. Miners earn TAO emissions based on the biological plausibility and realism of their synthetic datasets.
Use cases were auto-generated from the GitHub repository and may not reflect the latest changes.
NIOME is a decentralized Bittensor subnet (SN55) for generating synthetic CRISPR experiment datasets focused on the HBB (beta-globin) gene locus. Miners generate biologically plausible CRISPR experimental designs conforming to task specifications, while validators evaluate submissions through a progressive multi-stage validation pipeline (structural, statistical, biophysical, cross-consistency, and distributional fidelity) and reward miners with TAO emissions based on dataset quality and biological realism.
NIOME operates as a distributed system with four primary flows: (1) Task Generation: Backend generates CRISPR dataset generation tasks specifying mutations, cell types, and structural constraints, sent directly to validators; (2) Task Distribution: Validators broadcast identical tasks to all miners ensuring fair evaluation; (3) CRISPR Dataset Generation: Miners generate experimental design lists and upload to validator-provided S3 buckets via presigned URLs; (4) Multi-Stage Validation: Validators run a progressive validation pipeline on submissions through five increasing complexity stages, assign scores based on structural validity, statistical plausibility, biophysical simulation, cross-consistency, and distributional fidelity, and determine TAO emissions. Miners receive tasks via HTTP from validators, process them asynchronously (fire-and-forget pattern), and upload results. Validators maintain high uptime, execute multi-stage validation chains, manage S3 submissions, and produce consensus signals for the network.
The design prevents cheating by having miners submit only experimental designs while validators compute all biological outcomes.
This information was auto-extracted from the GitHub repository and may be incomplete or out of date. Always refer to the subnet's GitHub repository for the latest instructions.
The incentive mechanism operates through a multi-stage validation architecture: Stage 1 (Structural Gate) validates DNA syntax, PAM sequences, guide length, and mutation whitelist compliance; Stage 2 (Statistical Plausibility) evaluates GC content, spatial consistency, and structural realism; Stage 3 (Biophysical Simulation) simulates cut probability, repair pathways (HDR/NHEJ), and indel distributions using models like Azimuth/Rule Set 3 and inDelphi/FORECasT; Stage 4 (Cross-Consistency) checks learned biological relationships using ML models; Stage 5 (Distributional Fidelity) compares synthetic datasets to real CRISPR data distributions. Miners submit only experimental designs (guide sequences, targets, mutations) without predicted biological outcomes, preventing optimization shortcuts. Validators compute all outcomes, ensuring anti-cheating design.
The sections above were auto-generated from the GitHub repository and may not reflect the latest changes. View repository
NIOME is a decentralized AI subnet that enables privacy-safe genomic intelligence by replacing real human genomes with high-fidelity synthetic genomic profiles
FAQs were auto-generated from the GitHub repository and may not reflect the latest changes.
Holder & total-staked metrics as of 1h ago (hourly snapshot)
Up-to-date GPU generation recommended; 3rd party APIs unnecessary
1TB SSD storage recommended for dataset processing and validation pipeline; Docker required for PharmCAT; must maintain high uptime for scoring and consensus
Clone the NIOME subnet repository
git clone https://github.com/genomesio/subnet-niome.git && cd subnet-niome
Create and activate Python virtual environment
python3 -m venv venv && source venv/bin/activate
Install required Python packages
python3 -m pip install -r requirements.txt
Install Bittensor CLI for wallet management
python3 -m pip install bittensor-cli
Create coldkey (primary wallet for storing TAO)
btcli wallet new_coldkey --wallet.name your_coldkey
Create hotkey (miner identity for signing transactions)
btcli wallet new_hotkey --wallet.name your_coldkey --wallet.hotkey your_hotkey
Register hotkey to NIOME subnet (netuid 55) to obtain UID
btcli subnet register --netuid 55 --wallet.name your_coldkey --wallet.hotkey your_hotkey
Set Python path for project execution
export PYTHONPATH="$PYTHONPATH:$(pwd)"
Launch miner neuron (requires specifying port)
python neurons/miner.py --netuid 55 --subtensor.network finney --wallet.name your_coldkey --wallet.hotkey your_hotkey --axon.port your_port
Use pm2 or tmux to keep miner running persistently (validators reward only active, responsive miners)(optional)
pm2 start neurons/miner.py --name niome_miner -- --netuid 55 --subtensor.network finney --wallet.name your_coldkey --wallet.hotkey your_hotkey --axon.port your_port
Clone the NIOME subnet repository
git clone https://github.com/genomesio/subnet-niome.git && cd subnet-niome
Create and activate Python virtual environment
python3 -m venv venv && source venv/bin/activate
Install required Python packages
python3 -m pip install -r requirements.txt
Set up Docker repository
sudo apt-get update && sudo apt-get install -y ca-certificates curl gnupg && sudo install -m 0755 -d /etc/apt/keyrings && curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg
Add Docker APT source
echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu $(. /etc/os-release && echo $VERSION_CODENAME) stable" | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
Install Docker packages
sudo apt-get update && sudo apt-get install -y docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
Verify Docker installation
docker run --rm hello-world
Create coldkey and hotkey for validator registration
btcli wallet new_coldkey --wallet.name your_coldkey && btcli wallet new_hotkey --wallet.name your_coldkey --wallet.hotkey your_hotkey
Register validator hotkey to subnet 55
btcli subnet register --netuid 55 --wallet.name your_coldkey --wallet.hotkey your_hotkey
Make validator entrypoint script executable
chmod +x entrypoint.sh
Run validator interactively (prompts for wallet.name, wallet.hotkey, optionally wandb.api_key; sets up venv, installs dependencies, installs system tools (bwa, samtools, tabix, bcftools), starts validator as PM2 process, enables auto-restart on git updates)
./entrypoint.sh
