Skip to content

Overview

Architecture

vrAnalysis is organized around a few core concepts:

Sessions

A session represents a single experimental run. It's the central object that enables accesss to data for an experiment. In general, all more complicated analyses are built on top of session objects - in other words, the interface enabled by session is the bottleneck to all analysis of the package.

Each session contains:

  • Behavioral data (position, velocity, rewards, etc.)
  • Imaging data (calcium traces, ROIs, etc.)
  • Metadata (mouse name, date, session ID, etc.)

Sessions are represented by the B2Session class, which provides methods to load and access data. (B2Session is one example of a possible SessionData class, but since all of my work was done at B2, this is really the only one that's used. The other ones are for converting external data into a format that can be used by this package, which are explained elsewhere).

Database

The database tracks all sessions and their metadata. It's a simple interface that allows you to query your database easily and efficiently, and also to perform batch operations on sessions in the database. It uses Microsoft Access (.accdb) files by default, but can be adapted to other SQL databases.

The SessionDatabase class provides methods to:

  • Query sessions based on criteria
  • Add new sessions
  • Update session metadata
  • Track registration status and quality control flags

Processors

Processors transform session data into analysis-ready formats. For example:

  • SpkmapProcessor: Creates spatial maps of neural activity
  • I haven't built any others yet, maybe one day!

Tracking

Tracking identifies the same cells across multiple sessions. This enables longitudinal analysis of how individual cells change over time. This module is built with the Tracker class which is a nice wrapper around ROICaT tracking files. It will only work (or be relevant) if you've used ROICaT to track cells across sessions.

Registration

Registration is the process of preprocessing raw experimental data. This includes:

  • Loading behavioral data from Timeline files
  • Processing imaging data from suite2p outputs
  • Running OASIS deconvolution
  • Processing red cell annotations
  • Aligning data in time

Registration creates standardized data structures that can be used for analysis.

Data Flow

Raw Data (Timeline, vrControl, suite2p)
    ↓
Registration (B2Registration)
    ↓
Session Data (B2Session)
    ↓
Processing (Processors)
    ↓
Figures!

Directory Structure

Not sure where else to put this, but vrAnalysis expects data organized in an Alyx-style structure. This is the structure that is used by the alyx database and is generally a really nice way to organize sessions of data colleted on certain days from specific mice. 

localData/ # configure this in the `local_data_path()` function of `vrAnalysis/files.py`
├── mouse001/ # mouse name
│   ├── 2024-01-15/ # date of session 
│   │   ├── 001/ # session ID
│   │   │   ├── suite2p/ # where your suite2p output should go 
│   │   │   ├── onedata/ # where onedata will be stored after registration
│   │   │   └── ... # other files
│   │   └── 002/ # session ID
│   └── 2024-01-16/ # date of session
└── mouse002/ # mouse name