Reg. Allen Brain Observatory data download


Hi, I am using the allensdk to download certain ophys experiments of my interest. Since I am downloading the data for custom stimulus conditions the first time, it takes a good amount of time to download each experiment pertaining to my stimulus conditions, which is then cached in the BrainObservatoryCache manifest file.

It would be great to download for once, all of the experimental data for a brain area and hence I was wondering if you had access to bigger manifest files that have cached data for all experiments, organized by the area of the mouse visual cortex? I am currently working on data from the VISp area, and it would be very helpful if you could as well just share with me a manifest file with data for all VISp experiments cached in it. Thank you. :slight_smile:


Each experiment session has a unique NWB file, and downloading a lot of these can be time consuming. There are two ways you can approach this:

First, you can do your analysis on AWS. This way, there is no downloading – all of the data is kept in the cloud, you don’t have to spend anytime waiting for files to download, and you can do all of your analysis in the cloud.

Instructions for getting set up on AWS can be found here

Second, if you don’t want to work in the cloud, when you download a NWB file using the SDK you use the line:

>>> boc.get_ophys_experiment_data(ophys_experiment_id)

If you pre-select which experiment sessions you want, you can loop through and download them. For instance,

>>> exp = boc.get_ophys_experiments(targeted_structures=[‘VISp’])

Will give you a list of metadata for all of the experiment sessions in VISp. You could add a stimulus or session identifer to that call to even sub-select even further.

You can iterate over that output in order to download the NWB files for each of the selected experiments

>>> for i in len(exp):
>>> boc.get_ophys_experiment_data(exp[i][‘id’])


Hi @saskiad,
This is great, I will use the second technique. Thanks for the quick response :grinning: