Fig. 1.

A common workflow for the downstream analysis of scRNA-Seq data. The workflow includes the following seven steps: (i) quality control to remove low-quality cells that may add technical noise, which could obscure the real biological signals; (ii) normalisation and log-transformation; (iii) identification of the HVGs to reduce the dimensionality of the dataset by including only the most informative genes; (iv) standardisation of each gene to zero mean and unit variance; (v) dimensionality reduction generally obtained by applying PCA; (vi) clustering of the cells starting from the low-dimensional representation of the data that are used to annotate the obtained clusters (i.e., identification of known and putatively novel cell-types); (vii) data visualisation on the low-dimensional space generated by applying a non-linear approach (e.g., t-SNE or UMAP) on the reduced space calculated in step (v)