Single‑cell Intrinsic Conductance (IC) Analysis — Main Pipeline 🧠

Tools and notebooks for analyzing single‑cell intrinsic electrophysiology recordings (IO / RH / CH / SCH), computing features, and exporting publication‑quality figures with paired CSV data.

TL;DR 🚀

Install: pip install numpy pandas matplotlib seaborn scikit-learn
Data contract: one folder per cell, file may contain multiple blocks; channels fixed (Vm=0, Stim=1); units mV/pA/ms/MΩ/pF/nS; required tags: io_start/io_stop, rh_start/rh_stop, ch_start/ch_stop, sch_start/sch_stop; prominence=25, stim_correction=1000.
Run: open ./main_pipeline.ipynb, run the first time cell‑by‑cell, then Run All afterwards.
Outputs: every plot saved as SVG + PNG with an associated CSV of the plotted data, under analyzed/figures/<section>/ and analyzed/exports/.

What this pipeline does 📈

This pipeline indexes your single‑cell recordings, runs protocol‑specific analyses for:

IO (Input–Output): F–I curves, spike counts/frequency, rheobase‑related properties.
RH (Rheobase ramp): rheobase current per cell.
CH (Chirp, subthreshold): impedance vs frequency and peak power.
SCH (Short‑Chirp, suprathreshold): spike transfer vs frequency.

It then:

computes secondary properties (e.g., IO slope, F–I landmarks; AP waveform metrics),
supports QC/exclusion (e.g., fast spikers),
exports summary tables,
and saves every figure as PNG + SVG plus a CSV with the exact data used to draw the plot.

Install (one‑liner, no environments) 🔧

pip install numpy pandas matplotlib seaborn scikit-learn

The pipeline imports local modules kept in this repo (e.g., plots_with_stats.py, get_index.py, batch_process.py, io_secondary_properties.py). Make sure these files are present alongside the notebook/scripts.

Data contract (important) 🧾

Index & conditions: the folder name is used to create the data index. The underscore _ separates conditions.
- Example: CellA_ctrl_drug1 → cell_id=CellA, conditions: ctrl, drug1.
One file per cell, but the file can contain multiple blocks.
Channels (fixed for all files):
- data channel (Vm) = 0
- stim channel (current) = 1
Units:
- membrane potential mV, current pA, time ms, resistance MΩ, capacitance pF, conductance nS
Comment tags (epoch boundaries):
- IO: io_start & io_stop
- Rheobase ramp: rh_start & rh_stop
- Chirp (subthreshold): ch_start & ch_stop
- Short‑chirp (suprathreshold): sch_start & sch_stop
Spike detection: prominence = 25 for all cells.
Stimulus correction: factor = 1000 applied to all stim traces.

Keep the channel order and units consistent across files to avoid downstream conversion issues.

Folder structure & paths 🗂️

You control paths via the top‑level settings in the notebook (e.g., main_path, analyzed_path). A typical layout:

<repo_root>/
├─ main_pipeline.ipynb          # ← entry point
├─ batch_process.py
├─ get_index.py
├─ io_secondary_properties.py
├─ plots_with_stats.py
├─ ...other helpers...
└─ <main_path>/                 # your raw/processed cell data root
   ├─ CellA_ctrl_drug1/
   │  └─ CellA.csv (or .npz)
   ├─ CellB_ctrl/
   │  └─ CellB.csv
   └─ analyzed/                 # (created by the pipeline)
      ├─ io_basic/              # per‑protocol features & summaries
      ├─ io_wave/
      ├─ rh/
      ├─ ch/
      ├─ sch/
      ├─ figures/               # all exported figures
      │  └─ <section>/<name>.png, .svg, .csv, .meta.json
      └─ exports/               # run metadata, exclusion lists, combined tables
         └─ run_metadata.json, excluded_cells.csv, ...

How to run the main notebook ▶️

Launch Jupyter and open ./main_pipeline.ipynb (at the repo root).
First run: execute top‑to‑bottom cell‑by‑cell to create output folders, index files, and validate paths/settings.
After the first run, you can Run All.
Inspect outputs under analyzed/figures/ (PNG+SVG+CSV per plot) and analyzed/exports/.

Notes

The indexing step writes index.csv under main_path and prints a summary of discovered events.
QC step writes analyzed/exports/excluded_cells.csv describing excluded cell IDs (e.g., fast spikers).

What the other scripts do 🧰

get_index.py — Builds a stimulus event index from your data root (derives cell_id and conditions from folder names; applies channel assumptions).
batch_process.py — Batch runner that executes analyses over all cells for the specified protocols (io, rh, ch, sch), applying stimulus correction = 1000 and spike prominence = 25.
io_secondary_properties.py — Computes secondary metrics (e.g., IO slope, firing‑rate landmarks) and AP waveform properties.
plots_with_stats.py — Plotting helpers and statistical overlays used by the notebook.
plots/… or plot_*.py (if present) — Additional figure builders for specific analyses.

All plotting entry points use the shared exporter so every figure includes SVG + PNG + CSV (and a small .meta.json).

Troubleshooting 🛠️

No cells detected: check main_path and that folder names match the contract (underscore‑separated conditions).
Missing tags: ensure each protocol has the proper *_start/*_stop markers.
Incorrect channels: verify Vm is channel 0 and Stim is channel 1.
Units off: confirm data are in mV/pA/ms and that stimulus correction = 1000 is applied once.

License 📜

Unless otherwise noted, this project is released under Apache‑2.0 (see LICENSE).

pantelisantonoudiou/single_cell_ic_analysis