.._Ephys multi-sess-units:
Ephys: following units across multiple sessions¶
By default, the Ephys pipeline offers zero guarantees about the consistency of unit identities across separate sessions, as this happens, if at all, on the level of clustering, independent of the pipeline. Consequently, responsibility for ensuring that the units are actually correctly mapped from one session to the next remains with the researcher
There are three approaches by which you can maintain this important information
Option 1: multisession clustering¶
In this case, the clustering routine is applied simultaneously to multiple sessions, naturally yielding consistent, unique, cell identifiers.
The researcher is responsible for running their preferred Clustering routine.
Afterwards, the data can either be entered into the database via Python (see the helper notebook in the Ephys repository), or via the web GUI
Ephys -> Add Clustering -> Add Multi Session Clustering
Option 2: completely manual¶
In the process of clustering, you the user, adjust the identities attached to each unit to maintain a unique, consistent set of identifiers. When these clusterings are inserted as individual, single-session clusterings, the consistent identifiers are maintained within the pipeline. Therefore, session_1
unit_1
should be exactly the same physical unit as session_2
unit_1
, and if unit_1
does not appear in session_3
, then session_3
should not have a unit_1
label.
Option 3: Completely manual, spreadsheet ingestion¶
Option 2 functionally requires the researcher to maintain their own records to keep using consistent, unique, identifiers. This approach accepts one format of that data to be tracked in the pipeline.
This approach is currently not supported by the website interface, and requires use of a Jupyter notebook to proceed. (see the helper notebook in the Ephys repository)
The data can be provided as an Excel spreadsheet
Please contact Simon Ball if you require any assistance getting this to work, as you may need to be given explicit database permissions.
In this case, the clustering of each session does NOT use unique identifiers. For example, unit_1
in session_1
may be a completely different physical unit to unit_1
in session_2
. They are instead tied together by some unique identifier that indicates that session_1
unit_1
is the same as session_2
unit_7
and session_3
unit_4
.
These unique identifiers are stored in the table ephys.MultiSessUnit
and ephys.MultiSessUnit.MemberUnit
.