scVI Data integration
this task permit to perform data integration and clusters identification
- After data filtration using "Data filtration" task we proceed to data normalization and then data integration using scVI (single-cell Variational Inference). In fact scVI is a method based on a conditional variational autoencoder [Lopez et al., 2018] available in the scvi-tools package [Gayoso et al., 2022].
- This step permit the elimination of batch effect and then clusters determination based on leiden algorithm [Traag et al. (2018)], an improved version of the Louvain algorithm [Blondel et al. (2008)]
- As input, this task will take the combined_filtered.h5ad file generated by "Data filtration" task.
Input
combined_filtered_matrices
this file contain h5ad file generated by Data filtration task
Output
integrated_data
This file contain integrated data
clusters identification
This file contain the different identified clusters
integrated data statistical informations
This table stores some statistical informations related to integrated data
Configuration
clusters_resolution
A parameter controling the granularity of the clustering algorithm, influencing the number and size of the resulting clusters
float0.5