Figure 7. | Demixed principal component analysis of neural population data

Open accessCopyright infoDownload PDF

Demixed principal component analysis of neural population data

Figure 7.

Affiliation details

Champalimaud Centre for the Unknown, Portugal; École Normale Supérieure, France; Centre for Integrative Neuroscience, University of Tübingen, Germany; Wake Forest University School of Medicine, United States; Cold Spring Harbor Laboratory, United States; Universidad Nacional Autónoma de México, Mexico; El Colegio Nacional, Mexico; Harvard University, United States
Figure 7.
Download figureOpen in new tabFigure 7. Encoder weights for the leading dPCA components across the neural population.

(a) Distributions of encoder weights for the 15 leading dPCA components across the neural population, in each of the four datasets. Each subplot shows 15 probability density curves, one curve per component (bin width 0.005). The distribution corresponding to the first component is highlighted in red. (b) Clustering of neurons by density peaks (Rodriguez and Laio, 2014). For each dataset we took the first 15 dPCA components, and then ran the clustering algorithm in the 15-dimensional space of encoding weights. The clustering algorithm works in two steps: first, it computes a local density for each point (i.e., for each neuron), using a Gaussian kernel with σ2=0.01. Second, for each point it finds the minimal distance to a point with higher local density (if there is no such point, then the distance to the furthest point is taken). Each subplot shows local density on the horizontal axis plotted against distance to the next point with higher density on the vertical axis; each dot corresponds to one of the N neurons. Cluster centres are characterized by high local density and large distance to the point of even higher density; they should appear as outliers in the upper-right corner of the plot (see Rodriguez and Laio, 2014, for details). In each case, there is only one such outlier (bigger dot), indicating a single cluster.

DOI: http://dx.doi.org/10.7554/eLife.10989.010