MEA data analysis

Spontaneous network activity was recorded for 15 min every day using the Axion Biosystems Maestro MEA at 37°C. We collected raw data and a spike list file, which contains the spike data for each recording. These files were used downstream to extract spike, burst, and network data, using code adapted from sjemea package created by Stephen Eglen (Eglen et al. 2014). At the end of the experiment, activity data were inspected to remove inactive electrodes and wells. We determined the parameters for detecting neuronal bursts and network events based on published reports and experimentation (McConnell et al. 2012; Mack et al. 2014). In order for an electrode to be considered active, we required that at least five spikes per minute were recorded. Wells in which fewer than 16 electrodes were active for >30% of the days of recording were considered inactive and removed from analyses. For synchronous network events, at least 16 electrodes (25% of the total in a well) were required to participate in a network event in order for the network event to qualify as a network spike or burst. Events with less participating electrodes were filtered. By using the adapted code, activity variables were calculated (Supplemental Methods). To analyze data over time, the values for each well between DIV8 and DIV16 were combined and a MWU test was performed. The labels on each well were then shuffled and permuted 10,000 times to create 10,000 data sets that were tested for significance using a MWU test. The P-value of the actual data set was then compared to the distribution of the P-values from the permuted data sets to determine whether differences between the groups were significantly different from random. For additional information pertaining to MEA data analysis and statistical testing, please see the Supplemental Methods.