Prerequisites (for running manuscript codes)

To run LIBRA pipeline form R or Python source code or any other metric generated in the manuscript the following environmental settings are required:

• Run R3_requirements.R under R 3.5.2 or higher R 3.X.X for automatically install all dependencies required before using LIBRA.

• Run R4_requirements.R under R 4.0.3 or higher R 4.X.X for automatically install all dependencies required before using LIBRA.

-Following libraries are not supported under R 3.X.X environment: Seurat_4.0.0, MOFA2_1.0.1.

To run LIBRA fine-tune pipeline generated in the manuscript the following environmental settings are required:

-Please install the following Python libraries under Python v3.7.1 or higher: scanpy_1.5.0, scvi_0.8.1 (for totalVI), anndata_0.7.5, pandas_1.3.4, numpy_1.18.1, scipy_1.7.1, keras_2.7.0 and multiprocessing_2_6_2.

Versatility:

LIBRA manuscript code was develop in R but feel free to use "rpy2" Python library (https://rpy2.github.io/) for running LIBRA on R snippet through Python console otherwise if your preprocessing was performed using Python but you are interested in running libra in its R implementation you are able to move it to R by using "reticulate" package (https://rstudio.github.io/reticulate/).

Usage (manuscript codes)

• LIBRA pipeline is made easy to be run especially for any Seurat package user.
• The code is executed/stored in Seurat R objects, this allows the user to benefit from the long ecosystem of functions and structures present in Seurat.
• Either Seurat3 in R 3.X.X environment or Seurat4 in R 4.X.X environment can be used by hand of LIBRA.
• The valid input for LIBRA is any pair of omic matrices assigning the cell information in the rows and the feature information in the columns.
• Easiest way of running LIBRA analysis is though sc-Libra python package. Package documentation is online available using "Read the Docs" platform.

Getting Started with LIBRA (manuscript codes)

Basic vignettes:

• Model training and integration/prediction vignette for a quick example.
• PPJI preservation metric computation vignette for a quick example.

Vignettes repository:

• All vignettes can be found here.