This repository contains the raw data from chronic recordings in the visual cortex of three mice. The data were analyzed and published in Steinmetz, Aydin, Lebedeva, Okun, Pachitariu et al. "Neuropixels 2.0: A miniaturized high-density probe for stable, long-term brain recordings", Science 2021 (doi.org/10.1126/science.abf4588).The data pertain to Figure 4 of that paper, where neurons are characterized by their responses to a battery of 112 natural images. The fact that neurons retain the same preferences for natural images over tens of days strongly suggests that the neurons themselves are the same across days.The folder "Images" contains the 112 images that were presented to the mice.The data are from 3 mice:AL031 (“Mouse 3”), recorded with a 1-shank Neuropixels 2.0 probeAL032 (“Mouse 1”), recorded with a 4-shank Neuropixels 2.0 probeAL036 (“Mouse 2”), recorded with a 4-shank Neuropixels 2.0 probeAs detailed in the paper, mice were implanted in the left primary visual cortex, and the probe was cemented to the skull. During recording sessions, mice were head-fixed in front of 3 screens (left, center, right). Each stimulus was presented for 1 s, on the right and central screen, while the left screen was gray. In the interval between stimulus presentations (at least 2 s, as indicated in the files) all screens were gray. Recordings were made in SpikeGLX (https://billkarsh.github.io/SpikeGLX/) so once uncompressed, the data will be in .bin format.Each recording is contained in a folder named after the animal and the date of the recording. Each folder contains:A .cbin file and its accompanying .ch file. This is the compressed raw data. To decompress it, use the following package: https://github.com/int-brain-lab/mtscompA meta file, containing information about the recording settings.A file called stimIDs, containing the reference numbers of images that were presented (560 total; 5 presentations, 112 images each). Each stimID corresponds to an image in the "Images" folder.A file called stimTimes, containing the onset times of each stimulus (also 560 total).A folder called ks, containing the output of a spike sorting algorithm (pyKilosort; https://github.com/int-brain-lab/pykilosort) for that recording. This output can be inspected and curated using phy (https://github.com/cortex-lab/phy). To do so you will need to define a correct path to the data file in the params.py file in the ks folder.Example code to load the dataTo explore the data, we provide the Matlab function call_readRawDataChunk. Using this function, you can read and explore compressed data. To use the function, you will need the code in the repository: https://github.com/fangq/zmat (also available as Matlab Add-On https://uk.mathworks.com/matlabcentral/fileexchange/71434-zmat). Specify the file path, the starting point, and the duration of the chunk you want to explore at the top of the file, and happy exploration!