Discrepancy Between Average Firing Rate and Sweep Data

I am looking at long square sweep data of some spiny neurons in the Cell Types Database and noticed that some cells have a discrepancy between the average firing rate as shown in the cell_types_specimen_details.csv and the actual spikes shown in the sweep data.

From my understanding, the ef__avg_firing_rate reported in cell_types_specimen_details.csv should be the firing rate of a long square sweep with stimulus amplitude just above rheobase. However, this number reported for some cells are higher than the firing rates of even the long square sweeps with the highest firing rates.

For example, the human spiny neuron with specimen ID “531520401” has a firing rate of 146 according to the cell_types_specimen_details.csv, but the long square with the highest firing rate that I found (spike number 63) only has 9 spikes per second. Another example would be cell “537313600”, who only has one spike during the stimulation period in all sweeps that has a spike, but is shown to have 63 firing rate per cell_types_specimen_details.csv. This problem seems quite frequent among the spiny neurons.

Additionally, the cell_types_specimen_detials.csv also does not agree with the num_spikes from the get_ephys_sweeps() function on the CellTypesCache objects.

The “firing rate” and “ef__avg_firing_rate” reported on the web and in the specimen details CSV, respectively, are actually the inverse of the average inter-spike interval. So because those cells you point out only fire a few spikes, but with a very short interval, they have a high reported firing rate.

I’m not quite sure why that’s what’s being reported rather than the number of spikes divided by the stimulus duration - I think it may have been because we had many fewer values in an earlier iteration of the cell types database, and the average ISI better reflected the diversity of firing patterns across the population.

Apologies for the confusion!

1 Like