USNM 2P imaging data analysis pipeline

This repository hosts the code base to process and analyze 2-photon calcium imaging data acquired upon stimulation of different cortical regions by ultrasound.

Acquisition protocols & raw data

Bruker 2P imaging system

Raw data from the Bruker imaging system consist of individual 16-bit deep, 256-by-256 pixels grayscale TIF images of fluorescence data acquired at a sampling rate of 3.56 Hz and spatial resolution of 1.97 μm / pixel.

Each acquisition protocol (or run) consists of 16 sonication trials with an inter sonication interval of approximately 30s.

Each trial is divided in 2 cycles:

• The first cycle (indicated by an odd number) consists of 10 frames of the pre-stimulus interval of the trial (ca. 2.8s).
• The second cycle (indicated by an even number) consists of 90 frames of the peri and post-stimulus intervals of the trial (ca. 25.2s).

The ultrasound stimulus is delivered concurrently with the acquisition of the 10th frame of the first cycle (i.e. ca. at t = 2.8s).

For each acquisition protocol, the data is stored in a folder named after the specimen, stimulation and acquisition parameters:

foldername = <mouse_line>_<nframes_per_trial>frames_<PRF>Hz_<stim_duration>ms_<sampling_rate>Hz_<stim_amplitude>MPA_<stim_DC>DC-<run_ID>

Inside this folder, data is stored as single-frame TIF files named after the same pattern, together with unique cycle and frame identifers:

filename = <foldername>_Cycle<cycle_number>_Ch2_<frame_number>.ome.tif

This informative nomenclature is used as a way to store metadata associated with each experimental run.

Additionally, a file named <foldername>.xml, containing information about the microscope acquisition parameters for the run, is also stored in the data folder.

Bergamo 2P imaging system

TO COMPLETE

Analysis code base

Prerequisites

• Python 3.8+
• Conda environment manager

These can be downloaded from https://www.anaconda.com/products/individual.

Installation

• Create a new anaconda environment called usnm2p: conda create -n usnm2p python=3.8
• Activate this anaconda environment: conda activate usnm2p
• Clone this repository: git clone https://github.com/shohamlab/usnm2p.git
• Install the downloaded code base as an editable Python package: pip install -e ./usnm2p. This will install all required dependencies.
• Install the pre-built package for PyTables directly from conda: conda install pytables
• Move to the usnm2p subfolder inside the repository: cd usnm2p/usnm2p
• Create a user configuration file called config.py, and define in it:
  • a variable called dataroot indicating the path to the root directory continaing all the data folders to be analyzed.
  • a dictionary called default_dataset indicating parameters of the default dataset to analyze, with the following keys:
    • mouseline: mouse line
    • expdate: experiment date
    • mouseid: mouse number
    • region: brain region
• Save the configuration file in the repository top folder. You're all set!

Usage

Always start by activating the usnm2p anaconda environment: conda activate usnm2p

Notebooks

The notebooks subfolder contains several analysis notebooks:

• dataset_analysis.ipynb: analysis pipeline for a single dataset (date, mouse, region): from fluorescence movies to ROI-specific statistics.
• mouseline_analysis.ipynb: analysis pipeline for processed data from all datasets of a given mouse line
• main_analysis.ipynb: to run the main parameter-dependency analysis across mouse lines
• spatial_offset_analysis.ipynb: to run the analysis of datasets pertaining to the effect of spatial offsets
• buzzer_analysis.ipynb: to run the analysis of datasets pertaining to the effects of ultrasound vs. noise

Command-line scripts

The scripts subfolder contains several command-line scripts to run preprocess/analyze data: Several command-line scripts can also be run from the terminal to execute analysis notebooks over specific subsets of datasets:

• preprocess_bruker_data.py: preprocess raw datasets from Bruker 2P imaging system
• run_dataset.py: analyze one/several single dataset(s) from the command line (runs the dataset_analysis.ipynb notebook in the background).
• run_mouseline.py: analyze one/several mouseline(s) from the command line (runs the mouseline_analysis.ipynb notebook in the background).

To access command line options, type in: python <script_name> -h

Authors & contributors

• Data acquired by various members of the Shoham Lab at New York University.
• Code written and maintained by Theo Lemaire