TL;DR
This paper introduces SCOM, a novel computational framework and database (SCOMdb) to analyze "ncRNA synergistic competition" across 31 cancer types. They found that cancers share common "hub" ncRNAs rather than specific interaction patterns, and these ncRNAs are implicated in drug resistance, tumor subtyping, and immune regulation, suggesting their potential as biomarkers.
What Problem Does This Paper Address?
While non-coding RNAs (ncRNAs) are known to be crucial regulators of gene expression and play significant roles in cancer development, the complex interplay between multiple ncRNAs, particularly in the context of competing endogenous RNA (ceRNA) networks, remains poorly understood. Specifically, the paper addresses the challenge of identifying and characterizing "synergistic competition" among ncRNAs, where multiple ncRNAs collectively compete for microRNAs (miRNAs), influencing gene expression in a coordinated manner. Understanding these complex interactions is vital for uncovering new biomarkers and therapeutic targets for various cancers.
Methods
The authors developed a framework called SCOM (Synergistic COMpetition) to identify ncRNA synergistic competition. The process involves two main steps:
- Predicting ncRNA-related ceRNA networks: SCOM extends the SPONGE method by integrating gene expression data (miRNAs, lncRNAs, pseudogenes, mRNAs) from the Cancer Genome Atlas (TCGA) and known miRNA-target interactions from multiple databases. It uses three criteria (significant miRNA sharing, positive correlation, and sensitive correlation) to predict ceRNA interactions.
- Inferring ncRNA synergistic competition networks: For each tumor type, SCOM applies three criteria to identify synergistic competition between pairs of ncRNAs: (i) significant synergistic competition with mRNAs, (ii) significant positive correlation between the ncRNAs, and (iii) significant sensitive correlation conditioned on the competing mRNAs.
Network topological analysis (degree, path length, density) and similarity calculations were used to characterize these networks. Hub ncRNAs were identified using a Poisson distribution cumulative probability.
Datasets
The primary dataset used is the transcriptomics data from The Cancer Genome Atlas (TCGA) project, comprising 8001 matched tumor samples across 31 tumor types. This dataset includes expression data for 17,212 ncRNAs (894 miRNAs, 4366 lncRNAs, and 11,952 pseudogenes) and 19,068 mRNAs. Prior information on miRNA-target interactions was integrated from five databases: miRTarBase v9.0, TarBase v8.0, LncBase v2.0, NPInter 4.0, and ENCORI.
Key Insights and Ideas
- Shared Hubs, Not Shared Interactions: Cancers tend to share common "hub" ncRNAs that play central roles in synergistic competition, rather than sharing the exact same complex interaction networks. This suggests a conserved core regulatory mechanism across different tumor types.
- ncRNAs in Drug Resistance: Synergistic competition ncRNAs are significantly enriched in pathways related to drug resistance, particularly for drugs like cisplatin, L-685458, and LBW242. This highlights their potential role in treatment failure and as targets for overcoming resistance.
- Molecular Subtyping Power: Synergistic competition ncRNAs demonstrate a strong ability to distinguish molecular subtypes within various cancers, outperforming methods relying solely on hub ncRNAs or baseline approaches.
- Immune System Connection: These ncRNAs are significantly involved in immune regulation, being associated with numerous immune-related pathways and immune cells, suggesting a link between cancer progression and immune responses.
Key Results
- The SCOM framework was applied to 31 cancer types, revealing a landscape of ncRNA synergistic competition.
- Pseudogene-related ceRNA networks showed a significantly larger number of interactions than lncRNA-related networks.
- A substantial portion of ncRNA synergistic competition networks exhibit scale-free and small-world network properties.
- Conserved ncRNA-related ceRNA regulations were found to be low across the 31 tumors, indicating significant heterogeneity.
- Synergistic competition ncRNAs showed high performance (e.g., accuracy > 0.70) in classifying BRCA subtypes and distinguishing high-risk from low-risk groups in various cancers.
- Enrichment analysis confirmed that conserved synergistic competition ncRNAs and hub ncRNAs are significantly associated with diseases, cell markers, and cancer hallmarks.
- The study identified specific drugs (cisplatin, L-685458, LBW242) whose resistance is linked to ncRNA synergistic competition networks.
Limitations and Future Work
- Experimental Validation: The authors acknowledge that while their framework provides significant biological insights, the identified synergistic competition ncRNAs need further experimental validation to confirm their role as carcinogenic biomarkers.
- Power Law Distribution: Not all ncRNA synergistic competition networks strictly follow a power-law distribution, suggesting that in some tumors, ncRNAs might interact more equally.
- Stoichiometry: The paper mentions that the stoichiometry of miRNAs and ncRNAs is a vital factor for predicting ceRNA interactions, but this aspect was not fully incorporated into their current prediction model. Future work could leverage deeper understanding of ceRNA mechanisms to improve prediction accuracy.
- Single-Cell Level: The current SCOM framework analyzes bulk RNA sequencing data. The authors suggest that with the advancement of single-cell and spatial RNA sequencing technologies, SCOM could be extended to explore ncRNA synergistic competition at the single-cell level, potentially uncovering more carcinogenic biomarkers.
- Broader ncRNA Classes: While the study focused on lncRNAs and pseudogenes, SCOM can be extended to investigate other ncRNA types like circRNAs.
Summary updated