SiCmiR Atlas: Single-Cell miRNA Landscapes Reveals Hub-miRNA and Network Signatures in Human Cancers

TL;DR

SiCmiR is a neural network-based method that predicts single-cell miRNA expression from gene expression data, enabling the creation of a comprehensive atlas of single-cell miRNA landscapes in human cancers.

Contribution

The paper introduces SiCmiR, a novel neural network model that predicts miRNA profiles from limited gene data and constructs the first extensive single-cell miRNA atlas.

Findings

SiCmiR achieves state-of-the-art accuracy in predicting miRNA expression across multiple cancer types.

The SiCmiR-Atlas contains data from over 9 million cells across 726 cell types, enabling detailed miRNA analysis.

The resource facilitates biomarker discovery and understanding of miRNA networks in single-cell contexts.

Abstract

microRNA are pivotal post-transcriptional regulators whose single-cell behavior has remained largely inaccessible owing to technical barriers in single-cell small-RNA profiling. We present SiCmiR, a two-layer neural network that predicts miRNA expression profile from only 977 LINCS L1000 landmark genes reducing sensitivity to dropout of single-cell RNA-seq data. Proof-of-concept analyses illustrate how SiCmiR can uncover candidate hub-miRNAs in bulk-seq cell lines and hepatocellular carcinoma, scRNA-seq pancreatic ductal carcinoma and ACTH-secreting pituitary adenoma and extracellular-vesicle-mediated crosstalk in glioblastoma. Trained on 6462 TCGA paired miRNA-mRNA samples, SiCmiR attains state-of-the-art accuracy on held-out cancers and generalizes to unseen cancer types, drug perturbations and scRNA-seq. We next constructed SiCmiR-Atlas, containing 632 public datasets, 9.36 million cells, 726 cell types, which is the first dedicated database of single-cell mature miRNA expression--providing interactive visualization, biomarker identification and cell-type-resolved miRNA-target networks. SiCmiR transforms bulk-derived statistical power into a single-cell view of miRNA biology and provides a community resource SiCmiR Atlas for biomarker discovery. SiCmiR Atlas is avilable at https://awi.cuhk.edu.cn/~SiCmiR/.