Numerous tools have been recently developed to predict disease phenotypes using single-cell RNA sequencing (RNA-seq) data. CloudPred is an end-to-end differentiable learning algorithm coupled with a biologically informed mixture model, originally tested on lupus data. This study extends CloudPred's applications to breast cancer disease phenotype prediction to test its robustness and applicability on untested and unrelated biological data. When applying a breast cancer single-cell RNA seq dataset, CloudPred achieved an area under the ROC curve (AUC) of 1 in predicting cancer status and performed better than a linear and Deepset models.
@article{arxiv.2402.11289,
title = {Predicting Breast Cancer Phenotypes from Single-cell RNA-seq Data Using CloudPred},
author = {Hossein Moghimianavval and Baharan Meghdadi and Tasmine Clement and Man I Wu},
journal= {arXiv preprint arXiv:2402.11289},
year = {2024}
}