Fig. 7: Information contents extracted by SCA analysis using a TF-based latent space.
From: Sparsely-connected autoencoder (SCA) for single cell RNAseq data mining

a QCC. b QCM. c QCM/QCC plot, where only cluster 7 shows, for the majority of the cells, both QCC and QCM greater than 0.5. d COMET analysis of SCA latent space. SOX5 was detected as first top ranked gene specific for cluster 7, using as input for COMET the latent space frequency table. Input counts table for SCA analysis is made by raw counts.