scLINE: A multi-network integration framework based on network embedding for representation of single-cell RNA-seq data.
J Biomed Inform
; 122: 103899, 2021 10.
Article
en En
| MEDLINE
| ID: mdl-34481921
ABSTRACT
Single-cell RNA sequencing (scRNA-seq) is fast becoming a powerful technology that revolutionizes biomedical studies related to development, immunology and cancer by providing genome-scale transcriptional profiles at unprecedented throughput and resolution. However, due to the low capture rate and frequent drop-out events in the sequencing process, scRNA-seq data suffer from extremely high sparsity and variability, challenging the data analysis. Here we proposed a novel method called scLINE for learning low dimensional representations of scRNA-seq data. scLINE is based on the network embedding model that jointly considers multiple gene-gene interaction networks, facilitating the incorporation of prior biological knowledge for signal extraction. We comprehensively evaluated scLINE on eight single-cell datasets. Results show that scLINE achieved comparable or higher performance than competing methods, including PCA, t-SNE and Isomap, in terms of internal validation metrics and clustering accuracy. The low dimensional representations learned by scLINE are effective for downstream single-cell analysis, such as visualization, clustering and cell typing. We have implemented scLINE as an easy-to-use R package, which can be incorporated in other existing scRNA-seq analysis pipelines or tools for data preprocessing.
Palabras clave
Texto completo:
1
Banco de datos:
MEDLINE
Asunto principal:
Redes Reguladoras de Genes
/
Análisis de la Célula Individual
Idioma:
En
Revista:
J Biomed Inform
Asunto de la revista:
INFORMATICA MEDICA
Año:
2021
Tipo del documento:
Article
País de afiliación:
China