'Mini analysis' misrepresents changes in synaptic properties due to incomplete event detection.

Patch-clamp recording of miniature postsynaptic currents (mPSCs, or 'minis') is used extensively to investigate the functional properties of synapses. With this approach, spontaneous synaptic transmission events are recorded in an attempt to determine quantal synaptic parameters or the effect of synaptic manipulations. However, at the majority of brain synapses these events are small, with many undetectable due to recording noise. The effects of incomplete detection were well appreciated in the early years of synaptic physiology analysis, but appear to be increasingly forgotten. Here we sought to characterise the consequences of incomplete detection on the interpretability of mini analysis, using simulated mPSC data to give full control over event parameters. We demonstrate that commonly reported measures such as mean event amplitude and frequency, are misrepresented by the loss of undetected events. Probabilistic loss of small events results in detected event amplitude distributions that appear biologically complete, yet do not reflect the underlying synaptic properties. With both simulated and experimental datasets, we demonstrate that specific changes in event amplitude are primarily detected as changes in frequency, compromising classical biological interpretations. To facilitate more robust data analysis and interpretation, we detail a means for experimental estimation of the event detection limit and provide practical recommendations for data analysis. Together, our study highlights how mini analysis is prone to falsely reporting synaptic changes, raising awareness of these considerations, and provides a framework for more robust data analysis and interpretation. KEY POINTS: 'Mini analysis' (patch-clamp recording of miniature synaptic currents, mPSCs) is widely used to assess synaptic function, relying on detection of spontaneous synaptic events. Detection of mPSC events is almost inevitably incomplete, as event amplitudes are close to the level of recording noise - a limitation that was well recognised in earlier literature but is often overlooked today. Using in silico simulated datasets, this study characterises how incomplete detection distorts reported parameters and the distributions of detected events. These effects can routinely compromise biological interpretation of mPSC data, particularly the interpretation of amplitude and frequency changes. We present a method for experimental estimation of the detection limit and make practical recommendations for maximally careful interpretation of mini data.