Downloading 30-60 minutes of RAW neuropixels data

How To
1

I am new to the Allen institute and would like to download 30-60 minutes (only a small amount - less than 5GB) of RAW Neuropixels data. Do you know how to easily download a small section of it?

2

Hi there, and thanks for the question! All of the raw data from our Neuropixels experiments is available as an AWS public dataset. Instructions on how to access it are available here: Use the Allen Brain Observatory – Visual Coding on AWS · AllenInstitute/AllenSDK Wiki · GitHub

Since each probe has 384 channels sampled at 30 kHz, 30 minutes of data will be around 40 GB. What are you planning to do with the data? I may be able to send you a smaller snippet if you only need 5 GB or so.

3

Thank you for the fast reply - I did not expect the files to be so densely packed - (my calculations give me about 3843000030*60 = 20,736,000,000 -order of 10s of GBs - so seems about right)

I would still appreciate it if you send me a 5GB snippet of data - what method do you want to use?

4

Here’s 5 GB of raw data for 384 channels from one probe: continuous.dat - Google Drive

You can load it in Python using:

import numpy as np
fname = '/path/to/file'
data = np.memmap(fname, dtype='int16')
data = np.reshape(data, (data.size//384, 384))
5

Thank you for the data! May I ask about some meta data - which part of the brain is it recorded from and what are the values representative of (i.e how do I convert values to millivolts?)

6

This probe passes through thalamus, hippocampus, and cortex. To convert to millivolts, just multiply the values by 0.195.