neurocomputers 0.2.0

Analysis engine for brain organoid data — 57 modules, from spike sorting to Organoid IQ

NeuroBridge API

The analysis engine for biological neural networks.

57 analysis modules · 125 API endpoints · One-click full report on any organoid recording.

Built for researchers working with brain organoids on FinalSpark, Cortical Labs, and university MEA platforms.

Live: neurocomputers.io | API Docs: api.neurocomputers.io/docs

---

Quick Start

# Clone
git clone https://github.com/Luckyguybiz/neurobridge-api.git
cd neurobridge-api

# Setup
python3 -m venv venv && source venv/bin/activate
pip install -r requirements.txt

# Run
python main.py
# → http://localhost:8847/docs (Swagger UI)

Generate synthetic data and analyze (against live API):

# Generate 30s of synthetic spike data
DSID=$(curl -s -X POST "https://api.neurocomputers.io/api/generate?duration=30" \
  | python3 -c "import sys, json; print(json.load(sys.stdin)['dataset_id'])")

# Run a cheap analysis first
curl "https://api.neurocomputers.io/api/analysis/$DSID/summary"

# Get NCI Score (Network Complexity Index)
curl "https://api.neurocomputers.io/api/analysis/$DSID/iq"
# → {"iq_score": 51.0, "grade": "C", ...}

# Full report (runs everything — takes a while)
curl "https://api.neurocomputers.io/api/analysis/$DSID/full-report"

Load the built-in FinalSpark 118-hour recording:

DSID=$(curl -s -X POST \
  "https://api.neurocomputers.io/api/load-local?filename=SpikeDataToShare_fs437data.csv&sampling_rate=437" \
  | python3 -c "import sys, json; print(json.load(sys.stdin)['dataset_id'])")

# 2.6M spikes, 32 channels, 4 organoids, 5 days
curl "https://api.neurocomputers.io/api/analysis/$DSID/summary"

Upload your own data:

curl -X POST "https://api.neurocomputers.io/api/upload" \
  -F "file=@my_spikes.csv"

Supported formats: CSV, HDF5, Parquet, JSON, NWB. 100 MB max per file.

Operational notes

• Rate limits: 60 req/min regular endpoints, 5 req/min for heavy analyses (bursts/connectivity/cross-correlation/transfer-entropy/full-report). Per-IP.
• Per-analysis timeout: 45 s. Slower computations return 504 Gateway Timeout with a readable message.
• Concurrent heavy computations: 1 at a time (single CPU semaphore). Others queue.
• Dataset capacity: 2 concurrent datasets in RAM; oldest is evicted and returns 410 Gone on subsequent access.
• Auto-subsampling: heavy endpoints cap input to 10K–20K spikes for pairwise O(N²) work. Responses include _subsampled: true when this happens.

---

Analysis Modules

Standard Neuroscience (8 modules)

Module	Endpoint	Description
Spike Analysis	/analysis/{id}/firing-rates	Spike detection, ISI, firing rates, amplitudes
Spike Sorting	/analysis/{id}/spike-sorting	PCA + K-means / HDBSCAN clustering
Burst Detection	/analysis/{id}/bursts	Network bursts, single-channel, profiles
Connectivity	/analysis/{id}/connectivity	Co-firing graph, degree, clustering
Cross-Correlation	/analysis/{id}/cross-correlation	Pairwise correlograms
Transfer Entropy	/analysis/{id}/transfer-entropy	Directed information flow
Statistics	/analysis/{id}/summary	Full summary, quality, temporal dynamics
Spectral	/analysis/{id}/power-spectrum	PSD, frequency bands, coherence

Novel Analysis (unique to NeuroBridge)

Module	Endpoint	What it does
Organoid IQ	/analysis/{id}/iq	Composite intelligence score (0-100) across 6 dimensions
STDP Mapping	/analysis/{id}/stdp	Spike-timing dependent plasticity curves
Learning Detection	/analysis/{id}/learning	Temporal changes in plasticity = learning episodes
Attractor Landscape	/analysis/{id}/attractors	Memory as dynamical attractors (Hopfield theory)
Phase Transitions	/analysis/{id}/phase-transitions	Neural reorganization moments + optimal stim timing
Predictive Coding	/analysis/{id}/predictive-coding	Does the organoid minimize prediction error?
Weight Inference	/analysis/{id}/weights	Infer synaptic connectome from spike timing
Weight Tracking	/analysis/{id}/weight-tracking	Watch learning happen in real-time
Neural Replay	/analysis/{id}/replay	Memory consolidation during "rest" periods
Reservoir Computing	/analysis/{id}/memory-capacity	Memory capacity benchmark
Causal Emergence	/analysis/{id}/emergence	Integrated information (Phi)
Multi-Timescale	/analysis/{id}/multiscale	Complexity at 12 timescales
Fingerprinting	/analysis/{id}/fingerprint	Unique organoid identity signature
Sonification	/analysis/{id}/sonify	Neural activity → audio WAV
Health Monitor	/analysis/{id}/health	Organoid viability assessment
Predictions	/analysis/{id}/predict/bursts	Burst probability forecast
Anomaly Detection	/analysis/{id}/anomalies	Isolation Forest on neural features
State Classification	/analysis/{id}/states	Resting / active / bursting states

Discovery Analysis (10 modules)

Module	Endpoint	What it does
Sleep-Wake	/analysis/{id}/sleep-wake	UP/DOWN state detection, slow-wave oscillations
Habituation	/analysis/{id}/habituation	Response decay — simplest form of learning
Metastability	/analysis/{id}/metastability	Kuramoto synchronization, brain-like state switching
Information Flow	/analysis/{id}/information-flow	Granger causality, hub detection
Network Motifs	/analysis/{id}/motifs	3-node subgraph patterns vs random
Energy Landscape	/analysis/{id}/energy-landscape	Ising model, attractor basins
Consciousness	/analysis/{id}/consciousness	PCI + Phi + recurrence composite score
Comparative	/analysis/{id}/comparative	vs cortical slice, C. elegans, fruit fly, mouse hippocampus
Turing Test	/analysis/{id}/turing-test	Can you distinguish from Poisson/LIF?
Welfare	/analysis/{id}/welfare	Health + suffering detection + recommendations

Learning & Memory (4 modules)

Module	Endpoint	What it does
Catastrophic Forgetting	/analysis/{id}/forgetting	Do early patterns survive over time?
Transfer Learning	/analysis/{id}/transfer	Does learning task A help task B?
Consolidation	/analysis/{id}/consolidation	Offline memory consolidation events
Channel Capacity	/analysis/{id}/channel-capacity	Multi-bit information capacity

Connectomics (5 endpoints)

Module	Endpoint	What it does
Full Connectome	/analysis/{id}/connectome	Weighted adjacency, communities, modularity
Graph Theory	/analysis/{id}/graph-theory	Rich-club, small-world, efficiency, centrality
Effective Connectivity	/analysis/{id}/effective-connectivity	Directed causal connections
Topology	/analysis/{id}/topology	Betti numbers, persistent homology

Experiments (6 endpoints)

Module	Endpoint	What it does
Pong	POST /experiments/{id}/pong/simulate	DishBrain Pong simulation
Logic Gates	POST /experiments/{id}/logic/benchmark	AND, OR, XOR, NAND benchmark
Vowels	POST /experiments/{id}/vowels/simulate	Brainoware 240-vowel classification
Memory Tests	/experiments/{id}/memory-tests	Working, short-term, long-term, associative
Closed-Loop	POST /experiments/{id}/closed-loop/simulate	DishBrain mode simulation
Architecture Search	POST /analysis/{id}/architecture-search	NAS for optimal stimulation

Full Report

# Run ALL analyses in one call (~5 seconds)
curl "http://localhost:8847/api/analysis/{id}/full-report"

Returns 21+ analysis results in a single JSON response.

---

Data Formats

Input

• CSV: columns time, electrode, amplitude
• HDF5: FinalSpark format or generic arrays
• Parquet: columnar format
• JSON: array of objects or nested structure
• NWB: Neurodata Without Borders

Column names are auto-detected. FinalSpark electrode indexing (modulo 32) is handled automatically.

FinalSpark Compatibility

• Sampling rate: 30 kHz (configurable)
• Spike detection: 6× median(σ) threshold (Section 4.2 of Jordan et al. 2024)
• MEA: 4 × 8 electrodes = 32 channels
• Electrode index > 32 → modulo 32 applied automatically

---

Tech Stack

• Python 3.12 + FastAPI
• NumPy / SciPy / Pandas — core computation
• scikit-learn — ML (Isolation Forest, KMeans, Ridge, PCA)
• h5py — HDF5 file support

---

API Reference

Data Management

POST   /api/upload              Upload spike data file
POST   /api/generate            Generate synthetic data
GET    /api/datasets             List loaded datasets
GET    /api/datasets/{id}        Dataset info
GET    /api/datasets/{id}/spikes Spike data with filters

Analysis

GET    /api/analysis/{id}/summary           Full statistics
GET    /api/analysis/{id}/quality            Data quality metrics
GET    /api/analysis/{id}/firing-rates       Time-binned rates
GET    /api/analysis/{id}/isi                Inter-spike intervals
GET    /api/analysis/{id}/amplitudes         Amplitude distributions
GET    /api/analysis/{id}/temporal           Temporal dynamics
POST   /api/analysis/{id}/spike-sorting      Spike sorting
GET    /api/analysis/{id}/bursts             Network bursts
GET    /api/analysis/{id}/connectivity       Connectivity graph
GET    /api/analysis/{id}/cross-correlation  Correlograms
GET    /api/analysis/{id}/transfer-entropy   Directed information
GET    /api/analysis/{id}/entropy            Shannon entropy
GET    /api/analysis/{id}/mutual-information Mutual information
GET    /api/analysis/{id}/complexity         LZ complexity
GET    /api/analysis/{id}/power-spectrum     Power spectral density
GET    /api/analysis/{id}/coherence          Spectral coherence
GET    /api/analysis/{id}/avalanches         Criticality
GET    /api/analysis/{id}/digital-twin/fit   LIF model fitting
POST   /api/analysis/{id}/digital-twin/simulate  Simulate twin
GET    /api/analysis/{id}/anomalies          Anomaly detection
GET    /api/analysis/{id}/states             State classification
GET    /api/analysis/{id}/pca                PCA embedding
GET    /api/analysis/{id}/stdp               STDP plasticity
GET    /api/analysis/{id}/learning           Learning episodes
GET    /api/analysis/{id}/iq                 Organoid IQ
GET    /api/analysis/{id}/predict/firing-rates  Rate forecast
GET    /api/analysis/{id}/predict/bursts     Burst probability
GET    /api/analysis/{id}/health             Health assessment
GET    /api/analysis/{id}/replay             Neural replay
GET    /api/analysis/{id}/sequences          Neural circuits
GET    /api/analysis/{id}/memory-capacity    Reservoir MC
GET    /api/analysis/{id}/nonlinearity       Computation benchmark
GET    /api/analysis/{id}/fingerprint        Organoid fingerprint
GET    /api/analysis/{id}/sonify             Audio sonification
GET    /api/analysis/{id}/rhythms            Rhythmic analysis
GET    /api/analysis/{id}/emergence          Phi / causal emergence
GET    /api/analysis/{id}/attractors         Attractor landscape
GET    /api/analysis/{id}/state-space        State space geometry
GET    /api/analysis/{id}/phase-transitions  Phase transitions
GET    /api/analysis/{id}/predictive-coding  Predictive coding
GET    /api/analysis/{id}/weights            Synaptic weights
GET    /api/analysis/{id}/weight-tracking    Weight changes
GET    /api/analysis/{id}/multiscale         Multi-timescale
GET    /api/analysis/{id}/full-report        ALL analyses

Export

GET    /api/export/{id}/csv    Download as CSV
GET    /api/export/{id}/json   Download as JSON

---

License

MIT

---

NeuroBridge — Biocomputing-as-a-Service Platform