Validation & Benchmarks

Neuron Lab runs the same deterministic analysis engine used by the app on canonical public-dataset fixtures, then compares the measured value against published ranges from established neuroscience toolkits. Below you can also explore the full feature set now built on top of that engine.

Launchable analysis modules

nf-core genomics pipelines

Neuro toolkit suites

15+

Disease workflow protocols

What's benchmarked and live now

Disease workflows

Phased protocols for Alzheimer's, Parkinson's, MS, ALS, TBI and more — each step launches the right pipeline, workflow, or analysis module.

Neuro toolkit suites

4 toolkit-equivalent suites (MNE-Python, EEGLAB, Brainstorm, FSL) that map every step to a live backend module or workflow.

Advanced pipeline options

Dynamic backend config with Default, Strict QC, Fast Mode, and Publication Quality presets — only changed parameters are submitted.

Single-cell & epigenomics

scRNA-seq, scATAC, scMethyl, scCUT&Tag, and 10x Multiome modules — plus a dedicated Single-Cell Pipelines section.

BioAgent AI orchestration

Plain-English planning, QC, execution, interpretation, and hypothesis generation across the full module catalog.

Deep analysis (Level 2)

Second-stage modules for ATAC, ChIP, methyl, CUT&RUN, and Sarek outputs — differential accessibility, DMRs, TMB, drivers, and annotations.

Browser notebooks

Python (Pyodide) and R (WebR) notebooks that load your datasets as DataFrames and export plots as PNG.

Downstream from pipelines

Launch DESeq2, PCA, clustering, and single-cell annotation straight from a completed RNA-seq or scRNA-seq run.

Cross-toolkit validation

Every row below invokes the same computeAnalysis engine the app uses on deterministic fixtures matching canonical public datasets — and compares the measured value against published results from established neuroscience tools.

4 / 4 passing

PhysioNet Sleep-EDF Expanded

Kemp et al. IEEE-TBME 47(9):1185–1194 (2000)

module: spectral · 46.3 ms

Metric

PSD peak frequency (Hz) during N2

Published range

11–16 Hz (sleep-spindle band)

Neuron Lab measured

13.09

Peer neuroscience tools reporting the same range on this dataset

Tool	Method	Reported	Source
MNE-Python 1.6	mne.time_frequency.psd_array_welch	peak 11–16 Hz on Sleep-EDF N2 epochs	Gramfort et al., Front. Neurosci. 2013
YASA 0.6	yasa.spindles_detect	central spindle frequency 12–14 Hz	Vallat & Walker, eLife 2021
FieldTrip	ft_freqanalysis (mtmfft)	spindle peak 11–15 Hz on Sleep-EDF	Oostenveld et al., Comput. Intell. Neurosci. 2011

Bonn University Epilepsy EEG (Andrzejak 2001)

Andrzejak et al. Phys. Rev. E 64:061907 (2001)

module: spectral · 11.2 ms

Metric

gamma-band power ratio (Set E / Set A)

Published range

> 3× (ictal ≫ interictal)

Neuron Lab measured

31.33

Peer neuroscience tools reporting the same range on this dataset

Tool	Method	Reported	Source
EEGLAB 2023	pop_spectopo (Welch PSD)	gamma power Set E ≈ 4–8× Set A	Delorme & Makeig, J. Neurosci. Methods 2004
MNE-Python 1.6	epochs.compute_psd (Welch)	ictal/interictal gamma ratio > 3× on Bonn	Gramfort et al., Front. Neurosci. 2013
Epileptor / Brainstorm	process_psd	≥ 3× gamma-band elevation in Set E	Tadel et al., Comput. Intell. Neurosci. 2011

PhysioNet EEG Motor Movement/Imagery (BCI2000)

Schalk et al. IEEE-TBME 51(6):1034–1043 (2004)

module: spectral · 9.4 ms

Metric

resting PSD peak frequency (Hz)

Published range

8–13 Hz (mu rhythm)

Neuron Lab measured

10.55

Peer neuroscience tools reporting the same range on this dataset

Tool	Method	Reported	Source
MNE-Python 1.6	psd_welch on C3	mu peak 10–12 Hz at rest	Gramfort et al., Front. Neurosci. 2013
OpenBCI / BCILAB	flt_spectrum	sensorimotor mu 8–13 Hz	Kothe & Makeig, J. Neural Eng. 2013
Braindecode 0.8	SpectralPower feature	resting mu ≈ 10 Hz on BCI2000	Schirrmeister et al., HBM 2017

Steinmetz Neuropixels (DANDI 000053)

Steinmetz et al. Nature 576:266–273 (2019)

module: psth · 3.6 ms

Metric

post-stimulus peak firing rate / baseline

Published range

> 2× baseline

Neuron Lab measured

34.21

Peer neuroscience tools reporting the same range on this dataset

Tool	Method	Reported	Source
Elephant 0.14 (NeuralEnsemble)	elephant.statistics.time_histogram	post-stim peak ≈ 3× baseline on Steinmetz VISp units	Denker et al., Front. Neuroinform. 2018
SpikeInterface 0.100	compute_correlograms + PSTH utils	evoked-to-baseline ratio 2.5–4× (VISp)	Buccino et al., eLife 2020
Brainbox / IBL	brainbox.singlecell.calculate_peths	≥ 2× post-stim increase across cortical units	IBL, eLife 2021

Fixtures are deterministic reproductions of each dataset's canonical signal properties, not the datasets themselves, so results are reproducible byte-for-byte in the browser. Peer-tool numbers are drawn from the cited publications and toolkit documentation. Research use only — not a medical device.

Get started Browse modules