Identification of cancer subtypes from single-cell RNA-seq data using a consensus clustering method.

Area of Science:

• Computational biology
• Genomics
• Cancer research

Background:

• Human cancers exhibit intratumoral heterogeneity due to distinct molecular signatures.
• This heterogeneity complicates cancer diagnosis and treatment.
• Single-cell RNA sequencing (scRNA-seq) offers insights into cellular heterogeneity but presents computational challenges for high-dimensional data clustering.

Purpose of the Study:

• To develop a robust computational framework for cancer subtype identification using scRNA-seq data.
• To address the challenge of clustering high-dimensional, noisy scRNA-seq datasets.
• To improve the accuracy of cancer subtyping for enhanced diagnosis and treatment.

Main Methods:

• Introduction of a consensus clustering framework named conCluster.
• Utilizing an ensemble strategy to fuse multiple basic partitions into consensus clusters.
• Application to real cancer scRNA-seq datasets for subtype identification.

Main Results:

• conCluster demonstrates superior accuracy in detecting cancer subtypes compared to existing scRNA-seq clustering methods.
• Co-expression network analysis was performed on identified melanoma subtypes.
• Identified subtypes exhibit distinct gene co-expression networks and functional enrichment patterns.

Conclusions:

• The conCluster framework effectively identifies distinct cancer subtypes from scRNA-seq data.
• These subtypes possess unique gene co-expression profiles and functional characteristics.
• The findings contribute to a better understanding of tumor heterogeneity and personalized cancer therapy.