Single units with (signal-to-noise ratio>2) were detected off-line using SciWorks, and then manually discriminated and clustered using Offline Sorter (Plexon Inc., Dallas, TX).
For visualization, columns (features) were clustered using hierarchical Ward clustering and manually rearranged if necessary.
Array data were clustered using Cluster 3.0.
After experiments, we reanalyzed the waveforms using Plexon Offline Sorter and manually clustered the waveforms of SUA in principal component space; waveform groups were defined as SUA only if they formed a distinct, non-overlapping cluster in principal component space.
To do this, principal component (PC) analysis was used to determine the diversity of the image cube, convex geometry was used to find the spectrally pure pixels, and then these pixels were manually clustered in PC space.
Compounds retrieved from the screening collections were manually clustered.
The accuracy of the genotype call was manually evaluated for the misclassification of homozygous and heterozygous clusters using the software's clustering algorithm.
Thus, we performed a first step of ligand clustering using the LigandScout Ligand-set Clustering tool with default settings except for the cluster distance threshold that was manually adapted for each NRLiSt BDB dataset.
Cell clusters were manually defined using phase images and the irregular AOI tool and the resultant groups were segmented based on an empirically determined positive staining threshold.
Clusters were manually refined using K lustaviewa software.
