← research  /  PatchSeqML

PatchSeqML

Drop in a raw current-clamp recording and get the full intrinsic-electrophysiology feature set — AP shape, F–I curve, rheobase, input resistance, sag, adaptation — plus QC and an RS / FS / IB firing-type call, as a report. eFEL-grade, in seconds, no SDK.

checking backend…

Analyze a recording

The live analysis server is not active right now.
You can still:  ·  view an example report  ·  get the source on GitHub  ·  run it locally: python patchseqml.py analyze yourcell.abf

Why it's different

  • Turnkey. File-in → report-out. Stimulus protocol auto-detected — no manual epoch tagging.
  • Complete. The full intrinsic set in one pass, a superset of Allen IPFX's AP-only scope.
  • Validated. Every feature cross-checked sweep-by-sweep against eFEL (100% agreement) + QC + sanity checks. Tested on Allen human neurons.
  • Frictionless. Pure CPU numpy/scipy — no GPU, no Allen SDK, no license.
AP threshold / peak / amplitude half-width up/downstroke phase-plane F–I gain rheobase input resistance membrane τ sag ratio/τ fast/slow AHP ISI CV adaptation rebound

Who it's for

  • Ion-channel & neurophysiology labs running step current-clamp protocols.
  • Patch-seq researchers needing fast, reproducible ephys feature vectors to pair with scRNA-seq / morphology.
  • Channelopathy & drug-screening groups comparing intrinsic excitability across genotypes or compounds.
  • Trainees who need reference-correct features without learning a feature-extraction SDK first.
Positioned as the turnkey, validated layer between raw recordings and reference libraries (eFEL, Allen IPFX) — and a free, scriptable alternative to GUI suites (pCLAMP/Clampfit, AxoGraph). Recordings are processed server-side and nothing is stored.  ·  github.com/gianyrox/patchseqml  ·  example report