everythingevolves
commented
2 years ago This problem is likely to require a more complicated solution that identifying a single, new time constant as would be suitable for other Cre lines. In the case of Gad2, we're labeling a combination of different interneuron subclasses with very different calcium buffering properties. As a result, when imaging in layer 2/3, for example, VIP, SST, and PV neurons will all be present in a single FOV. This means that the right solution will likely involve having (at least) two possible time constants that can be chosen from in order to assign one time constant to each ROI rather than a single time constant that is assigned to all ROIs in the FOV. However, this will require additional software infrastructure to classify each ROI as fast (e.g., putative VIP or SST neuron) versus slow (e.g., putative PV neuron), then pass those time constants to event detection as a list with one element for each ROI.
An alternative solution the requires less effort would be to choose a single intermediate value, or a single value that corresponds to the fast time constant such that VIP and SST neurons have good results whereas PV neurons will not be as good...but given that PV neurons have slow calcium dynamics that make GCaMP fluorescence look more like an integrated signal rather than easily separable (i.e., de-convolvable) events, we might be willing to accept a sub-optimal solution for PV neurons since no solution is likely to be perfect for PV neurons anyway due to this inability to deconvolve events perfectly for PV neurons.
DowntonCrabby
matchings
Description of needed information In order for LIMS event detection to run it needs a decay time constant for the GCaMP signal. however, we currently do not have these values for the new GAD2 reporter lines (Ai195 and Ai210). We have implemented a stop-gap solution (we just used the decay time for Pvalb:ai162) in order to allow LIMS processing to proceed but this is not a permanent solution.
Related issues No current issue
Describe the solution you'd like We would like to meet at the QC Stakeholders meeting to discuss this issue and discuss our options moving forward.
Screenshots or example plots NA
Scope This will impact ANY project using these new reporter lines (including Learning & mFISH and v1omFISH).
Additional context & analysis Adding @everythingevolves to this issue