# 2025 - Autism

This is the dataset associated with the publication [A common computational and neural anomaly across mouse models of autism](https://www.nature.com/articles/s41593-025-01965-8).

Electrophysiology, behavioral and video data were collected in mice performing the [IBL task](https://pubmed.ncbi.nlm.nih.gov/34011433/). 
Animals were either wildtype, or from one of 3 different autism models:
- _B6.129P2-Fmr1tm1Cgr/J_ JAX Strain #003025
- _B6.129(Cg)-Cntnap2tm2Pele/J_ JAX Strain #028635
- _B6.129-Shank3tm2Gfng/J_ JAX Strain #017688

The following data repository contains intermediate results and code to reproduce the analysis: https://osf.io/fap2s/ and https://osf.io/fap2s/wiki/home/. 

Please note that different quality standards were applied in the [article](https://doi.org/10.1038/s41593-025-01965-8) than what is presented in this page. For example, the IBL does not use a presence ratio to assess the quality of its good spike sorting units.

## Notes
The dataset has also been published via the [OSF platform](https://osf.io/fap2s/wiki/home/). However, downloading the dataset through OSF is not recommended.
We recommend you download the data via ONE.

## Overview of the Data
We have released data from 364 Neuropixel recordings, referred to as probe insertions. Those were obtained with 62 genetically modified subjects performing the IBL task.
As output of spike-sorting, there are 129,313 units; of which 22,008 are passing IBL automated quality control.

| strain   |   n_subjects | n_recordings | n_units | n_good_units |
|:---------|-------------:|-------------:|--------:|-------------:|
| Cntnap2  |           16 |           85 |  31,082 |        4,211 |
| Fmr1     |           15 |          105 |  34,401 |        6,215 |
| Shank3t  |           16 |           75 |  31,679 |        5,423 |
| Wildtype |           15 |           99 |  32,151 |        6,159 |
| Total    |           62 |          364 | 129,313 |       22,008 |

## Data structure and download
The organisation of the data follows the standard IBL data structure.

Here is a minimal example of how to download the data for one of the insertions using `ibllib`

```python
import pandas as pd
from brainbox.io.one import SpikeSortingLoader, SessionLoader
from one.api import ONE

one = ONE(base_url='https://openalyx.internationalbrainlab.org')
insertions = one.search_insertions(project='angelaki_mouseASD')
ssl = SpikeSortingLoader(one=one, pid=str(insertions[0]))
sl = SessionLoader(one=one, eid=ssl.eid)
sl.load_trials()
spikes, clusters, channels = ssl.load_spike_sorting()
df_clusters = pd.DataFrame(ssl.merge_clusters(spikes, clusters, channels))
print(df_clusters)
print(sl.trials)
```

To go further with more advanced examples, see:

* [These instructions](https://int-brain-lab.github.io/iblenv/notebooks_external/data_structure.html) to download an example dataset for one session, and get familiarised with the data structure
* [These instructions](https://int-brain-lab.github.io/iblenv/notebooks_external/data_download.html) to learn how to use the ONE-api to search and download the released datasets
* [These instructions](https://int-brain-lab.github.io/iblenv/loading_examples.html) to get familiarised with specific data loading functions

Note:

* The tag associated to this release is `2025_Q3_Noel_et_al_Autism`
* The project associated to this release is `angelaki_mouseASD`

## How to cite this dataset
If you are using this dataset for your research please cite:
- the paper [A common computational and neural anomaly across mouse models of autism](https://doi.org/10.1038/s41593-025-01965-8)
- if you download the data via ONE (recommended), cite the DOI: [10.6084/m9.figshare.30024880](https://doi.org/10.6084/m9.figshare.30024880)
  
- if you download the data via OSF (not recommended), please use the OSF DOI.

## Data release notes and changelog

### 2025-08-29: initial release

Possible future evolutions:
- Spike sorting re-run with iblsorter or newer algorithm to have better yield and better pre-processing
- Perfom new alignments of the brain region labels with regards to the ephys signatures using an automatic tool

#### Excluded sessions

11 sessions were excluded as the synchronisation stream from the bpod couldn't be found in the FPGA. As such the behaviour information couldn't be extracted to a clock synced to the electrophysiology.


#### Video data: QC report
We have only included video data where we could confidently align the frames timing to the main experiment clock.

##### Quality of uppermost channels
What we believe to be a failing Neuropixel headstage translates in the loss of 10-15 of the uppermost channels of the probes. This affects many recordings.
Make sure to perform an anomaly detection using [ibl-neuropixel](https://github.com/int-brain-lab/ibl-neuropixel) or [spike interface](https://spikeinterface.readthedocs.io/en/stable/)


#### Electrophysiology: QC report.

364 insertions were retained for the release, 102 were excluded for the reasons below.

##### CRITICAL: Missing Histology tracing (17 probes)
The following insertions do not have any tracing available and won't be recoverable.

##### ERROR: Missing spike sorting (36 probes)
The following insertions do not have any spike sorting available, they are not part of the current release but may be part of a future release.

##### ERROR: Missing alignments (49 probes)
Here we have spike sorting and histology tracing, but the channels haven´t  been aligned. We are not releasing those datasets for the time being.
Those would be good candidates for a future revision if the ephys atlas task force devises a tool to perform automatic alignments.