An extensive evaluation of single-cell RNA-Seq contrastive learning generative networks for intrinsic cell types distribution estimation (2)

This repository includes the pre-trained encoders used in the paper "An extensive evaluation of single-cell RNA-Seq contrastive learning generative networks for intrinsic cell types distribution estimation". Each ZIP file is named after the scRNA-Seq dataset that is used to train five contrastive learning methods (i.e. AF-RCL, Sup-GsRCL, Sup-RGMRCL-5000, Self-GsRCL, and Self-RGMRCL-3000). The experiments are run using 5-fold cross-validation settings and model selection is conducted using two metrics (i.e. ARI and NMI). Each pretrained encoder is named as follows: [Method name]--[Dataset name]--[Fold]--[Evaluation metric]_checkpoint.pt. For more details on how to load the pretrained encoders and reproduce the results, please check our GitHub repo at https://github.com/ibrahimsaggaf/scRCL-G.